HLTH 2120 Week 5 Data Quality Audit Example

Sixty Admission Weights, Thirty-Four Defects: A Data Quality Audit Sorted by Where in the Process Each Problem Entered

Student Name

Walden University

HLTH 2120: Health Informatics

Instructor Name

Month Day, Year

Sixty Admission Weights, Thirty-Four Defects: A Data Quality Audit Sorted by Where in the Process Each Problem Entered

The Records and the Question

The class was given an extract of 60 adult admissions to a medical-surgical unit over one month, with each patient's first recorded weight, its unit, the method of measurement, when it was charted, when the first dose calculated from weight was given and any weight on file from an earlier stay. The extract is a teaching data set built to resemble real records; it contains no real patients. The assignment asks what is wrong with these records. This audit answers in two steps. First, it counts the defects using a published framework, so the categories are not invented for the occasion. Second, it sorts every defect by the point in the process where it entered the record. The second step is the more useful one, because it turns a list of problems into an argument about how the capture of weight was designed.

Step 1: Counting the Defects

I used the harmonized data quality framework proposed by Kahn and colleagues, which groups data quality checks into conformance (do values follow the expected format and rules), completeness (are values present) and plausibility (are values believable given what else is known) (Kahn et al., 2016). Table 1 shows what the audit found.

Table 1

Defects Found in 60 Admission Weights, by Category

Note. Some records had more than one defect. The provenance and timing rows go beyond the three core categories of the framework and were added because they matter for dosing.

Step 2: Where Each Defect Entered

Counting shows that 26 of 60 records, 43%, had at least one defect. That number alone suggests careless documentation. Sorting the defects by where they entered the process suggests something else.

At the emergency department. Nine of the 11 stated weights and both pound-for-kilogram errors were entered in the emergency department before the patient reached the unit. The emergency department's triage screen offers a single weight box whose unit defaults to pounds, while the inpatient flowsheet defaults to kilograms. The two conformance errors were both cases where a triage nurse typed a pound value and the interface converted it as if it were kilograms. These defects entered at a design boundary between two screens with different defaults.

At the admission flowsheet. The three blank method fields and the five copied values entered here. The flowsheet allows the method field to be skipped, and it offers a button to pull the most recent prior weight, which is faster than weighing a patient. Both are design choices that make the defect the path of least effort.

At the ordering step. The six records where the first weight came after the first weight-based dose share a pattern: an anticoagulant was ordered from the emergency department using the triage weight, then the unit recorded a measured weight later. The order set accepted the triage weight because a weight existed. The timing defect entered where the order set checked for any weight rather than a measured one.

Undetermined. The four missing weights, the three implausible changes and the two stated weights entered on the unit itself could not be traced to a design feature from the extract alone.

What the Sorting Shows

Of the 34 defects, 25 entered at one of three design points: the mismatch between emergency department and inpatient units, the skippable method field with its copy-forward button and the order set that accepts any weight. The other nine could not be linked to a design feature. The audit therefore supports a different conclusion from the raw count. The problem is less that staff document poorly and more that the capture of weight was designed so that the quickest path produced a defect.

This is consistent with what reviews of electronic record data quality report: data collected for one purpose, such as triage, are often reused for another, such as dosing, without anyone checking whether they are fit for the second use (Weiskopf & Weng, 2013). The triage weight is good enough for triage. It was never designed to drive an anticoagulant dose.

Sorting also changes who should act on the findings. A defect list sent to the unit manager would lead to reminders and education for nurses. The sorted list points instead to the informatics team, which controls the screen defaults, the copy-forward button and the order set logic, and to the emergency department, which owns the triage screen. Education would still help, but it would be working against a design that rewards the shortcut.

Recommendations, Linked to Where the Defects Entered

Each recommendation targets a point of entry rather than a category. At the emergency department boundary, set both screens to kilograms and show the unit next to every displayed weight, in line with medication safety guidance that weights be recorded in metric units only (Institute for Safe Medication Practices, 2024). At the admission flowsheet, make the method field required and remove the copy-forward button for weight, or label copied values as copied. At the ordering step, have weight-based order sets check for a measured weight entered during the current visit and prompt for one when only a stated or triage weight exists, with an exception for emergencies.

These changes would be tested by repeating this audit on the next month's admissions. If the design argument is right, the stated-weight, conformance, copied-value and timing defects should fall sharply, while the undetermined defects should change little. That second prediction matters, because a fall in every category would suggest that staff were simply being more careful while watched, not that the design changes worked.

References

Institute for Safe Medication Practices. (2024). Targeted medication safety best practices for hospitals. https://www.ismp.org/guidelines/best-practices-hospitals

Kahn, M. G., Callahan, T. J., Barnard, J., Bauck, A. E., Brown, J., Davidson, B. N., Estiri, H., Goerg, C., Holve, E., Johnson, S. G., Liaw, S.-T., Hamilton-Lopez, M., Meeker, D., Ong, T. C., Ryan, P., Shang, N., Weiskopf, N. G., Weng, C., Zozus, M. N., & Schilling, L. (2016). A harmonized data quality assessment terminology and framework for the secondary use of electronic health record data. eGEMs (Generating Evidence & Methods to Improve Patient Outcomes), 4(1), Article 1244. https://doi.org/10.13063/2327-9214.1244

Weiskopf, N. G., & Weng, C. (2013). Methods and dimensions of electronic health record data quality assessment: Enabling reuse for clinical research. Journal of the American Medical Informatics Association, 20(1), 144-151. https://doi.org/10.1136/amiajnl-2011-000681