NR 542 Week 6 Dataset Analysis Example

Dataset Analysis: Twelve Months of Early ICU Transfers From Medical Floors, 118 of 3,940 Admissions, Reported With the Figure and With Everything That Had to Be Dropped, Assumed or Left Uncertain to Produce It

Student Name

College of Nursing, Chamberlain University

NR 542: Managing Data and Information

Instructor Name

Month Day, Year

Dataset Analysis: Twelve Months of Early ICU Transfers From Medical Floors, 118 of 3,940 Admissions, Reported With the Figure and With Everything That Had to Be Dropped, Assumed or Left Uncertain to Produce It

Two Ways to Be Wrong

There are two ways to be wrong about a dataset, and only one of them shows up in the total. The first is to calculate the figure incorrectly. The second is to calculate it correctly from data that do not mean what the reader thinks, because of who was excluded, what was assumed or how uncertain the result is. This analysis reports the 12-month rate of early moves from a medical ward to intensive care, for adult medicine patients who came in through the emergency department, and it reports with equal care what had to be dropped, assumed or left uncertain to produce it. It uses the rules in the reporting plan, revised after the data quality audit.

The Figure

In the 12 months from July to June, 3,940 encounters met the denominator rules. Of these, 118 patients left the ward for intensive care before their first day there was out, planned moves aside. The rate is 3.0%, and the interval around it runs from 2.5% to 3.5%. This is well below the 8% quoted at quality meetings and near the rate a large California system reported for its own ward admissions from the emergency department (Delgado et al., 2013), although that system's population and definition differ from ours.

Monthly counts ranged from 5 to 15, so monthly rates will vary substantially by chance. The report will emphasize the rolling 12-month rate.

The Other Half: What Was Dropped

Producing the figure required excluding encounters at several steps, and each exclusion shapes what the figure means.

Planned transfers. Sixty-one transfers within 24 hours were excluded because a procedure requiring intensive care monitoring was scheduled. Most were cardiac catheterizations and endoscopies. If they had been included, the rate would have been 4.5%.

Comfort care. Forty-four patients with comfort care orders signed before floor arrival were excluded. Few such patients are transferred to intensive care, so their exclusion slightly raises the rate among those remaining.

No recorded floor arrival. Eighty-seven emergency department admissions to hospital medicine had no movement into a medical-surgical unit. Most were admitted directly to the intensive care unit or step-down unit, which is appropriate. But 12 were patients who died or were discharged from the emergency department while boarding, and 9 had bed movement records the audit could not reconcile. These 21 are a small group whose outcomes the report cannot capture.

Step-down transfers. Seventy-four patients moved from a floor to a step-down unit within 24 hours. They are not in the numerator. If step-down transfers were counted as escalations, the rate of any early escalation would be 4.9%. The report will show this figure separately, because some readers will consider step-down transfer the same kind of event.

Reporting these exclusions is not a disclaimer added at the end. Each one changes what the figure describes, and a reader who wants a broader or narrower measure can see exactly how much it would move.

The Other Half: What Was Assumed

The figure assumes that bed movement times reflect real movement closely enough. The audit found a median lag of 18 minutes and occasional lags over two hours, so the true window for some patients was shorter than 24 hours. It assumes that the corrected unit table is right for the whole year, although 5 West changed use fourteen months ago, before the analysis period began. And it assumes that scheduled procedures identify planned transfers, although some planned transfers may have been arranged without a scheduled procedure. Operational record data carry these kinds of hidden assumptions because they were collected for care and billing rather than measurement (Hersh et al., 2013).

Where the Rate Differs

The rate was higher for patients arriving on the floor at night, during the night shift hours: 52 of 1,270, or 4.1% (95% confidence interval 3.0% to 5.2%), compared with 66 of 2,670, or 2.5% (1.9% to 3.1%), for daytime arrivals. The intervals barely touch, and the difference may reflect sicker patients arriving at night, fewer staff at night or chance. Patients admitted with pneumonia or sepsis accounted for 49 of the 118 transfers, consistent with published findings that respiratory conditions and sepsis are associated with early unplanned transfer (Delgado et al., 2013).

For the secondary question about warning signs, 45 of the 118 transferred patients, or 38%, had a National Early Warning Score of 5 or more in the four hours before leaving the emergency department, compared with about 9% of patients who were not transferred. The score was validated against exactly this outcome, among others, and separated patients who would need unplanned ICU care from those who would not with good accuracy (Smith et al., 2013). The finding suggests that warning signs were present for a substantial minority of transferred patients, but it does not show that the admission decision was wrong: a score of 5 is common and does not by itself indicate a need for intensive care.

What the Analysis Supports

The analysis supports three statements. The early transfer rate is about 3%, not 8%. It is higher at night, and the difference deserves investigation. And more than a third of transferred patients had elevated warning scores before leaving the emergency department, which suggests a subgroup whose placement could be reviewed. The analysis does not support a statement that admission decisions are poor, and the report to the director will say so in its first paragraph. The 45 patients with elevated scores are the obvious starting point for the clinical case review the director asked for, because data can point to cases but only clinicians can judge them.

The analysis will be repeated when the new transfer-order questions have three months of data, so that the reasons for early transfer can be set beside the rates. Until then, the case review of patients with elevated warning scores is the best available way to learn whether placement decisions could be improved, and the analysis has been shared with the two physicians who will lead that review, together with the list of patients and the definitions used to select them.

The full query set and this analysis are filed together.

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

Hersh, W. R., Weiner, M. G., Embi, P. J., Logan, J. R., Payne, P. R., Bernstam, E. V., Lehmann, H. P., Hripcsak, G., Hartzog, T. H., Cimino, J. J., & Saltz, J. H. (2013). Caveats for the use of operational electronic health record data in comparative effectiveness research. Medical Care, 51(8, Suppl. 3), S30-S37. https://doi.org/10.1097/MLR.0b013e31829b1dbd

Smith, G. B., Prytherch, D. R., Meredith, P., Schmidt, P. E., & Featherstone, P. I. (2013). The ability of the National Early Warning Score (NEWS) to discriminate patients at risk of early cardiac arrest, unanticipated intensive care unit admission, and death. Resuscitation, 84(4), 465-470. https://doi.org/10.1016/j.resuscitation.2012.12.016