NR 542 Week 5 Data Quality Audit Example

Data Quality Audit: Two Hundred Encounters Checked Against the Chart for the Early ICU Transfer Report, Six Problems Found, From a Preselected Dropdown to a Unit Still Coded as Something It Stopped Being Last Year, and What Each Does to the Figure

Student Name

College of Nursing, Chamberlain University

NR 542: Managing Data and Information

Instructor Name

Month Day, Year

Data Quality Audit: Two Hundred Encounters Checked Against the Chart for the Early ICU Transfer Report, Six Problems Found, From a Preselected Dropdown to a Unit Still Coded as Something It Stopped Being Last Year, and What Each Does to the Figure

Purpose and Method

Wrong numbers rarely come from wrong arithmetic. They come from a field that was optional, a dropdown whose first entry was left selected, and a patient who exists twice under two spellings. Before the first monthly early-transfer report goes out, this audit checks whether the fields the report depends on are accurate enough to support it.

The audit drew a random sample of 200 encounters from the report's candidate population for the previous 12 months: 150 from the denominator without a transfer and all 50 transfers in the numerator, since the numerator is small and errors there matter most. For each encounter, a second nurse informaticist compared the reporting database values with the patient's chart: the admission source in the emergency department note, the time of arrival on the floor in the first nursing note, the unit and its current use, the team in the admitting note, and any intensive care transfer in the transfer note. The checks map onto the quality dimensions Weiskopf and Weng (2013) set out for reusing record data: whether values are present, right, consistent across sources and up to date.

Problem 1: A Preselected Admission Source

The admission source field on the admission order defaults to "Emergency department." Clerks and physicians who admit patients directly from clinics or transfers from other hospitals sometimes leave the default in place. In the sample, 11 of 200 encounters, 5.5%, were coded as emergency department admissions but were direct admissions or transfers from other hospitals, according to the chart. Direct transfers from other hospitals are more likely to be sick, so their inclusion may raise the rate slightly. The fix is to use the emergency department encounter link, which exists only for true emergency department admissions, rather than the admission source field.

The second reviewer did not know which encounters were transfers when checking the denominator sample, so that expectations about sicker patients would not influence how charts were read. Each discrepancy was recorded with the field, the database value, the chart value and the likely cause.

Problem 2: A Unit Coded as What It Used to Be

One unit, 5 West, was converted from a medical-surgical unit to a step-down unit fourteen months ago. The unit type table still codes it as medical-surgical. Patients admitted directly to 5 West from the emergency department are therefore treated as floor admissions, and patients moving from a medical floor to 5 West are not counted as step-down transfers. In the sample, 9 encounters had 5 West as their first unit. This is a currency problem: the table was correct when written and has not kept up. The fix is to update the table and to add a check against the current unit list to each monthly run.

Problem 3: Bed Movement Times That Lag

Bed movement times are entered by a clerk or nurse when the move is recorded, not when it happens. Comparing them with the first nursing note on arrival, the median lag was 18 minutes, but in 7 of 200 encounters it exceeded two hours, usually at night. A lag at floor arrival delays time zero and shortens the effective 24-hour window. The audit found no case in which the lag changed whether a transfer counted, but two transfers occurred within 30 minutes of the 24-hour boundary. The report will note this limitation, and the collection design will consider automatic time stamps.

Problem 4: Duplicate Encounters

Before the record system upgrade two years ago, observation patients converted to inpatient status sometimes generated a second encounter. The audit found 3 such pairs in the sample, all before the upgrade. In each pair, the first encounter ended with the conversion, and the second began at the same moment without a floor arrival, because the patient did not move. The report's build sequence would drop the second encounter for lacking a floor arrival but would miss any intensive care transfer recorded on it. The fix is to link converted encounters in the build for periods before the upgrade.

Problem 5: Missing Team Assignment

Team assignment was missing at the time of floor arrival for 4 of 200 encounters, usually when a patient arrived on the floor before the admitting team was entered. The chart showed hospital medicine in all four. The fix is to use the first team assignment within four hours of floor arrival.

Problem 6: One Patient, Two Records

Two encounters in the sample belonged to a patient registered under two medical record numbers because of a misspelled surname. This does not affect an encounter-based measure directly, but it would affect any analysis of repeat transfers by patient. The duplicate has been referred to health information management for merging.

Effect on the Figure

The audit rebuilt the 12-month figure with each fix applied in turn. The rate under the original build sequence was 3.4%. Removing mislabeled direct admissions and transfers reduced it to 3.2%. Correcting the unit type for 5 West reduced it further to 3.0%, because some patients admitted to 5 West and transferred to intensive care had been counted as floor admissions. Linking duplicate encounters and filling team assignments made no measurable difference to the rate. The corrected figure is therefore about 3.0%, far below the 8% quoted at meetings and not far from the rate one large health system published (Delgado et al., 2013), although differences in definition limit that comparison.

What the Audit Changes

The report's build sequence has been revised to use the emergency department link, the updated unit table and linked encounters. Two new checks are added to each run. A data quality framework for secondary use of record data recommends documenting known limitations alongside any measure (Kahn et al., 2016), so the first report will include a one-paragraph summary of this audit. The audit will be repeated on a smaller sample each year, and sooner if the record system or unit structure changes.

A summary of the audit, including the effect of each fix, will be kept with the reporting plan so that the next analyst can see why each rule exists.

References

Delgado, M. K., Liu, V., Pines, J. M., Kipnis, P., Gardner, M. N., & Escobar, G. J. (2013). Risk factors for unplanned transfer to intensive care within 24 hours of admission from the emergency department in an integrated healthcare system. Journal of Hospital Medicine, 8(1), 13-19. https://doi.org/10.1002/jhm.1979

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, 4(1), Article 18. 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