Overcoming Clinical Data Problems in Quality Improvement Projects
When a workgroup team begins to focus on a operational or clinical quality improvement initiative, the excitement and anticipation of actually being able to access data is tangible. Hopes are high. Previously the challenge of bringing these types of projects to fruition has been data access. Clinical and operational managers have not been able to reach the level of data specificity they desired to achieve goals such as reducing 30-day heart failure readmissions or improving patient transfer time. Now, the team realizes that times have changed. They can arm themselves with data and accomplish something great!
As the project progresses, data quality issues frequently bring the team down from these euphoric heights into despair and finger pointing. The reality is that team members expect their enterprise data quality to be higher than it usually is. Analysis uncovers data discrepancies and shows the need to improve data capture. It can feel like another roadblock. During project kick-off meetings, we show a diagram nicknamed the Trough of Despair. We do this to warn clinicians of this natural event that occurs with any type of data discovery. It’s unavoidable; wherever clinical data exists, issues with that data will exist. The best way to limit the depth of despair is communication and teamwork.
For example, we were working with a clinical team to analyze cesarean section rates for first-time mothers pregnant with a single, full-term infant, where the baby’s head was in the vertex position—essentially someone who should have a normal vaginal delivery. As we gathered the data elements to calculate these rates, it became apparent that numerous individual data elements were missing. For instance, the number of pregnancies and the number of live births were document in different locations within the electronic health record (EHR). To make matters worse, the same data point documented in two different places by two different users could show two different values. Which value was the team to believe? Other data elements may not have been documented at all. Finally, the EHR was automatically adding a ‘1’ to the number of pregnancies column following delivery, thus increasing the chance for inaccurate numbers. The team felt overwhelmed. How do they continue their quality improvement project with all these new challenges? Looking at data capture is a good place to start.
Improving Clinical Data Capture
As a workgroup team reviews and analyzes their data, the discovery process begins. There may be missing data, inaccurate data, data that doesn’t provide enough detail or no data at all. The focus needs to be on how to improve data capture, even though the task may feel overwhelming.
In our example cesarean example above, missing data was another issue. Dilation values (a key determinant in deciding if cesarean sections are justified) were often omitted. When the workflow for obtaining this value was analyzed, we discovered that a key field to record dilation values was located at the end of a long flow sheet, and the clinicians often closed the flow sheet before getting to the end of the document.
Missing data may also result from multiple entry points, changes in data location, clinicians not understanding the importance of entering a specific piece of information or difficulty finding a specific data field. This may be remedied by changing a location, making a field mandatory (you are unable to move on with documentation until the field is completed) or providing staff education on the importance of a specific data element.
Inaccurate data most often occurs through errors in documentation. For instance, an incorrect value might be entered in a blood pressure field. It might also be a system calculation based on erroneous data—if the wrong birth year was entered, the patient will appear older or younger than they really are. Updated provider tables can also be a challenge, as can data entered as free text versus structured entries.
Finally, often there is not enough detail associated with the data. One of our client teams was trying to understand why there had been an increase in cesarean section rates. Clinicians were given two options to document the reason a section occurred—either maternal or fetal. Clearly, this was not enough detail for the work team to analyze processes and make recommendations for change. New data selections were approved; rationale was provided to clinicians and then implemented into the documentation system. Within 24 hours, the work group team was able to see the new data and began a more detailed analysis.
The trough of despair can seem ever widening! However, the data simply needed to be improved, and the team was equipped with the tools and processes to make that happen.
The good news is that by having the perseverance and commitment to analyze the data, or lack thereof, the team is given opportunities to work with the clinicians and the EHR documentation teams to align documentation to the workflows, and to teach the importance of accurate and meaningful documentation. The end results of this effort are to provide data that credibly reflects what has actually occurred, and most importantly, improved patient care.
What phases have you seen with clinical data discover? Have you experienced the elation of having access to new data? How about the trough of despair?