HINF 500 Week 5 Applying Data to an Administrative Decision Example

Did the Falls Pilot Work? Using Control Charts, Comparison Units and Published Evidence to Decide Whether to Spend $186,000 Expanding a Program to Eight Units

[Student Name]

University of Phoenix

HINF/500: Informatics for Health Administration

Week 5 Assignment

[Instructor Name]

[Date]

The hospital, units, fall data and costs are composites written for a model paper; research findings come from the sources listed.

The nurse manager of a 32-bed medical unit at a composite 310-bed community hospital asked the vice president of operations for $186,000 to expand a fall prevention program to all eight adult inpatient units. Her unit had piloted the program for nine months: a fall risk assessment linked to a tailored prevention plan, a bedside poster showing each patient's risks and interventions, hourly rounding for high-risk patients and a post-fall huddle after every fall. By her count, the unit had averaged 3.9 falls for every thousand days of inpatient care over the prior two years, and only 2.8 since the pilot began. The vice president wanted to know whether the drop was real. This paper describes how the data were used to decide.

Defining the Decision

The decision was specific: spend $186,000 in the first year, mostly for training time, bedside posters and screens and a part-time coordinator, to expand the program, or not. The question the data had to answer was whether the pilot had caused a real reduction in falls, large enough to justify the cost.

Sourcing and Checking the Data

Falls came from the incident reporting system and patient days from the census system, both extracted monthly for 33 months, 24 before the pilot and 9 during it, for all eight units. The analyst checked that every fall report had a unit and date and compared a sample of reports with nursing notes. One complication emerged: the hospital had changed incident reporting software two months into the pilot, which could affect how many falls were reported.

Why Averages Are Not Enough

A drop from 3.9 to 2.8 sounds meaningful, but monthly fall rates on a single unit vary widely by chance. In the baseline period, the unit's monthly rate ranged from 1.6 to 6.4. Comparing two averages cannot tell whether nine months of lower rates reflect a real change or ordinary variation.

Using a Control Chart

Statistical process control offers a better method. A control chart plots a measure over time with a center line and limits calculated from the data's own variation, allowing managers to separate common-cause variation, the everyday scatter any steady process produces, from special-cause variation, which means some real influence has entered (Benneyan, 2003). A systematic review found control charts applied across many health care settings and specialties to help teams manage and evaluate change, while noting that their value depends on correct application (Thor et al., 2007).

What the Chart Showed

The analyst built a chart for rates suited to counts per patient day, using the 24 baseline months to set the center line at 3.9 and the limits. No single pilot month fell outside the lower limit. But all nine pilot months fell below the center line, which satisfies a widely used shift rule requiring at least eight successive months below, or above, the baseline average. The chart did not show one dramatic month; it showed nine quiet ones in a row, which chance rarely produces.

Testing Rival Explanations

Three alternatives were checked. First, a hospital-wide improvement: the seven other units, charted the same way, showed no shift during the same months. Second, the reporting software change: total incident reports on the pilot unit, for all event types, did not fall after the change, and falls on the other units did not drop when they adopted the same software. Third, a change in patients: the unit's average age and share of patients over 75 were similar before and during the pilot.

The Outcome That Matters Most

Injurious falls, the result that matters most to patients and relatives, fell from 11 in the last nine baseline months to 5 during the pilot. The numbers are small, so the difference is suggestive rather than conclusive.

Outside Evidence

Published research supports the result. In a randomized trial in four hospitals, units using a fall prevention tool kit that linked risk assessment to tailored bedside interventions recorded 3.15 falls per thousand patient-days against 4.18 on usual-care units, with the largest effect among patients 65 and older, though fall injuries did not differ significantly (Dykes et al., 2010). The pilot's result is close to that trial's.

Why Not Wait for More Data?

The chief financial officer asked a fair question: why not run the pilot another year before deciding? Waiting has costs too. At the baseline rate, the seven other units would experience roughly 60 more falls in the coming year than they would with the program, if the pilot's effect holds. More months would narrow uncertainty only modestly, because the shift was already clear on the chart. And the program's main risk, that it works less well elsewhere, can only be tested by trying it elsewhere. A phased expansion answers both concerns: it acts on the evidence while generating new data before full commitment.

What the Numbers Cannot Show

The analysis could not show which parts of the program mattered most. Nurses on the pilot unit believed the post-fall huddles changed behavior more than the posters, while the manager credited hourly rounding. Because all parts started together, the data cannot separate them. The expansion will therefore keep all elements but survey staff and track compliance with each, so that a future decision about simplifying the program rests on more than opinion.

The Cost Side

At the pilot unit's rate of improvement, expansion to eight units would prevent roughly 60 falls a year. If even a modest share would have caused injuries requiring added treatment and longer stays, avoided costs could approach the program's cost, apart from the harm avoided for patients.

The Decision

The vice president approved expansion in two phases: four units in the first six months and the remaining three after a review, with $186,000 budgeted over the year. Phasing allows the hospital to check that results hold on units with different patients before full commitment. The first four units were chosen because their baseline fall rates were highest, where the program has the most room to help.

Monitoring

Each unit will have its own control chart, reviewed monthly by nurse managers and quarterly by the patient safety committee, along with falls with injury and compliance with the bedside plan. The chart's center line will be recalculated once each unit has enough program months.

Limits

The pilot was on one unit, with an enthusiastic manager. Results may be smaller elsewhere. The control chart shows that the rate changed, not why, so staff observations of which parts of the program mattered will be collected.

Conclusion

A before-and-after average would have given a misleading sense of certainty. A control chart showed a real shift, comparison units and a reporting check ruled out obvious rival explanations and a randomized trial supported the result. The data supported a phased expansion with monitoring, a decision that can be revisited if the charts change.

References

Benneyan, J. C. (2003). Statistical process control as a tool for research and healthcare improvement. Quality and Safety in Health Care, 12(6), 458-464. https://doi.org/10.1136/qhc.12.6.458

Dykes, P. C., Carroll, D. L., Hurley, A., Lipsitz, S., Benoit, A., Chang, F., Meltzer, S., Tsurikova, R., Zuyov, L., & Middleton, B. (2010). Fall prevention in acute care hospitals: A randomized trial. JAMA, 304(17), 1912-1918. https://doi.org/10.1001/jama.2010.1567

Thor, J., Lundberg, J., Ask, J., Olsson, J., Carli, C., Härenstam, K. P., & Brommels, M. (2007). Application of statistical process control in healthcare improvement: Systematic review. Quality and Safety in Health Care, 16(5), 387-399. https://doi.org/10.1136/qshc.2006.022194