Learning from errors: zooming in on healthcare incident data

Can we do more with the incident data that we collect? That was the subject of the study that Wouter van Kempen (student Msc. Occupational Health Psychology, at Leiden University) successfully completed in close cooperation with a respectable Dutch hospital, during his CGE internship. Read Wouter’s account on how they approached the research and what they learned from it.

Medical errors and how they are collected

It’s a common mistake: the hospital pharmacy receives a medication order for Oxynorm 5mg, a heavy painkiller similar to morphine. What is prepared for pick-up is Oxycontin 5mg, which, although similar in name and application, is a long-acting variant. The hospital software system can theoretically differentiate between the two, but small blisters of opiates do not have barcodes on them. As a result, scanning of the barcode for registration is not possible, therefore sidestepping of the verification process. In this case, potentially causing the patient to receive the wrong kind of medication.

This type of medication error and many others are collected in the incident-databases of hospitals and care facilities. This bulk of data is used for identifying trends and generating reports and tracking the quality of service in healthcare.

The project: can we do more with the incident data we collect?

An internship at CGE has been set up around the question: can we do more with the incident-data we collect? Can a technique or instrument be devised to visualize and improve medication safety?

To start this project, CGE offered a solid 5-day training covering BowtieXP, IncidentXP, BowtieServer and AuditXP, next came a flexible workspace and introduction to the business-contacts with subject-matter expertise.

The research approach

We started with a database from the hospital incident-reporting system, specifically on medication-errors. This resulted in general recommendations centered around ‘human errors’ and operational procedures. This however proved of little help to the people reporting incidents regarding medication errors and how their work can be improved: general recommendations have a limited impact.

Next, we arranged incident-data per department (e.g. main pharmacy, intensive care units, satellite divisions, etc.) and ranked the most common errors and incidents. The reported incidents gave some insights into what went wrong, such as: infusion-pump with medical residue, late or not delivered medication, wrong dosage or missing label, etc. This became the basis for further incident analysis: Tripod Beta incident diagrams.

With the information provided by the departments, we built preliminary Tripod incident diagrams to clarify:

  • Tripod Immediate Cause (an action or omission by a person or group of people that causes a Barrier to fail). This information was gathered from the database
  • Tripod Precondition (a ‘state of mind’ or emotional/cognitive climate by which the immediate cause becomes more likely). Information collection was interview-based
  • Tripod Underlying Cause (the organizational deficiency or anomaly on a ‘system level’). Information collection was interview-based

To verify and improve these diagrams with staff, we scheduled interviews with Hospital pharmacists and nurses about these incidents in a non-threatening way (i.e. “guide me through your process; what happened here?”) and this helped to clarify preconditions and underlying causes. From this point, we were able to place barriers in a Bowtie. A final feedback-session was planned with employees to proactively identify barrier strengths in a bowtie. Note: this couldn’t have been done without incident-data and prior investigation. In this final session, we invited two hospital divisions (employees of a Intensive Care unit and Satellite IC division) to identify barriers that work and barriers that did not, as well as barrier strengths and escalating factors. In doing so, we introduced barrier-thinking to employees in a constructive way.

What we learned

Data is a great way to start any investigation; if sorted properly and incident descriptions are clear, it can guide the user towards what’s happening and set the theme for interviews. Filling in the blanks has to be done with face-to-face contact in the form of interviews with department personnel. From this point, the barrier way of thinking is introduced and concluded with a Bowtie.
The conversations based on incident-data have provided key insights into the day-to-day processes of healthcare personnel, why certain incidents can happen and how certain risks were avoided for example by making use of protocol or technology.

Conclusion

Firstly, the approach presented here is a first step towards a framework that integrates quantitative incident-data and qualitative interviews, to visualize organizational past incidents and proceed with a Bowtie that identifies threats and consequences on a department-level in hospitals, because each operational level has their own distinct features and therefore specific risks involved.

Secondly, these incident-diagrams and Bowties can be incorporated into the departmental quality reviewing process. It is a way for employees start a conversation on common known risks (for example missing barcodes on blisters of medication) and a starting point to think about barriers and risks associated with performing daily operations; how these barriers are working and risks can be managed.

Thirdly, starting reactively with an incident-investigation and ending proactively with a bowtie provides employees an overview of functional barriers that are working well. This also fits into the safety 2.0 perspective: identifying how things go right, which is often the basis for how things can go wrong.

Find out more about this study

On the 20th of September, Wouter will be conducting a poster presentation with regards to his study at the AMC Symposium: “Incidenten in de Patiëntenzorg”. Curious to find out more about his research? We look forward to seeing you at this event.

2018-08-23T15:50:43+00:00Healthcare, News|

Leave A Comment