MHA 507 Week 6 Data Plotting and Information Analysis Example

Did the Float Pool Work? Plotting Thirty Months of Falls, Overtime and Burnout on Run Charts, Control Charts and a Scatter Plot

[Student Name]

University of Phoenix

MHA/507: Using Informatics in the Health Sector

Week 6 Assignment

[Instructor Name]

[Date]

The health system, its units and all data values are composites written for a model paper; methods and research findings come from the sources listed.

Six months after a composite four-hospital system placed a float pool of 30 registered nurses to reduce overtime on its twelve highest-overtime medical-surgical units, the chief nursing officer asked a direct question: had falls with injury really dropped, or was the system seeing a lucky stretch? The performance improvement manager answered with charts. This paper explains which charts she used, what they showed and how she presented them.

Questions Before Charts

The manager listed the questions first. Had falls with injury on the twelve units changed after the float pool began? Was any change larger than ordinary month-to-month variation? And across all units, did overtime and burnout move together as expected? Each question pointed to a different chart.

The Data

Monthly injurious falls and patient days for the twelve units came from the incident system and the census system for 24 months before the change and 6 months after, giving 30 monthly rates per 1,000 patient days. Unit-level overtime share and burnout pulse results for all 48 units came from payroll and the survey vendor for the most recent quarter.

Why Not a Bar Chart

A bar chart comparing the average rate before and after would hide the pattern over time. A single good month after the change could make the average look better by chance, and a trend that began before the change would be invisible. Data about change need to be plotted in time order.

Chart One: The Run Chart

The run chart plotted the 30 monthly rates in order, with a horizontal line at the median of the 24 baseline months, 0.60 injurious falls for every 1,000 patient days. A vertical line marked the month the float pool began. All six months after the change fell below the baseline median, joined by the final two months before it, as the pool was partly staffed in that period, making eight consecutive points below the median.

Reading the Run Chart

Standard run chart guidance counts a sustained streak, usually six months or longer, entirely above or entirely below the median as evidence that the process has shifted rather than wobbled. Eight straight months under the baseline median satisfied that test comfortably. Eight months in a row below the old median is unlikely to be luck; two good months would have been.

Chart Two: The Control Chart

Run charts detect shifts, but control charts add limits that show the expected range of ordinary variation. Statistical process control combines time-series analysis with graphics, giving decision makers faster and more understandable insight into data than methods that wait to aggregate months of results (Benneyan, 2003). Because falls are counts over varying patient days, the manager used a u-chart, which calculates limits that widen in months with fewer patient days.

What the Control Chart Showed

In the baseline, one month exceeded the upper limit at 1.12 falls per 1,000 patient days. Investigation traced it to a single unit during a flu surge with heavy agency staffing, a special cause. With the new process's data, the center line after the change was 0.41, and the eight straight months beneath the old center line counted as a shift under the chart's standard rules.

Evidence on Using These Charts

A systematic review of 57 empirical studies found statistical process control applied across many settings and specialties with 97 different variables, identified benefits, limitations and barriers and concluded that its value depends on correct and thoughtful application, including risk adjustment and data stratification (Thor et al., 2007).

Stratifying by Unit

Following that advice, the manager plotted the twelve units separately as small multiples. Ten showed lower rates after the change; two did not. Both were units where the float pool had filled fewer than half of open shifts because of their location, which pointed to a specific fix rather than a failed program.

Why Rates, Not Counts

The twelve units' patient days varied from month to month by as much as 15%, with lower census in summer and higher census in winter. Plotting raw counts of falls would have made busy winter months look worse simply because there were more patients to fall. Dividing by patient days put every month on the same footing, and the u-chart's limits reflected the fact that a rate based on fewer patient days is less certain than one based on more.

Checking the Data Before Plotting

Before drawing anything, the manager checked the data. Two months had missing patient days for one unit, which were recovered from the census system's backup reports. Three falls had been entered twice by different staff members, and the duplicates were removed. The definition of a fall with injury had not changed during the 30 months, so the series was comparable from start to finish.

Chart Three: The Scatter Plot

To test the relationship between workload and morale, the manager plotted each of the 48 units with overtime share on the horizontal axis and the proportion of staff reporting burnout on the vertical axis. The points rose from lower left to upper right, with a correlation of 0.58.

Reading the Scatter Plot

The scatter plot shows that overtime and burnout move together, not that one causes the other. Units with difficult patient populations might have both. Two outlying units, with high overtime but low burnout, were worth visiting to learn what protected their staff.

Designing for the Board

Research on displays of health data found that simple tables were understood well by many audiences, that consistent formats and contextual cues aided understanding and that less information often produced better comprehension (Hildon et al., 2012). The board therefore received one table and one chart: the u-chart, with the change annotated and a single sentence explaining what the limits mean.

The Board Table

The table has four rows, one for each measure, and three columns: baseline, the six months since the change and the target. Falls with injury went from 0.60 to 0.41 against a target of 0.45; overtime share from 11% to 6% against 5.5%; burnout from 48% to 42% against 40%; and registered nurse turnover, measured on an annualized basis, from 24% to 19%.

Presenting to Nurse Managers

Nurse managers received a different package from the board: their own unit's run chart, the system's u-chart and the scatter plot with their unit's point highlighted. Seeing their unit in context prompted questions the board would not ask, such as why one unit's falls clustered on night shift, and two managers asked for their data by shift, which the manager added to the next quarter's charts.

Design Choices

Every chart had a title stating its finding, labeled axes with units, a note on data sources and dates and no three-dimensional effects or decorative color. Red was used only to mark the special-cause month.

Limits

Six months is a short period, and falls could rise again in winter. The float pool began at the same time as a fall prevention refresher on two units, which could explain part of the change on those units. The scatter plot uses one quarter of survey data.

What Comes Next

The manager will keep the u-chart running monthly, recalculating limits only after a sustained shift is confirmed; fix float pool coverage on the two lagging units; and repeat the scatter plot each quarter to see whether overtime reductions are followed by lower burnout.

Conclusion

The chief nursing officer's question was whether the drop in falls was real. A run chart showed eight consecutive months below the old median, a control chart confirmed a shift and explained an earlier spike and stratified charts found two units where coverage fell short. A scatter plot linked overtime and burnout without claiming cause. Picking each chart for its question and reading the result with established rules turned monthly numbers into a decision the board could trust.

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

Hildon, Z., Allwood, D., & Black, N. (2012). Impact of format and content of visual display of data on comprehension, choice and preference: A systematic review. International Journal for Quality in Health Care, 24(1), 55-64. https://doi.org/10.1093/intqhc/mzr072

Thor, J., Lundberg, J., Ask, J., Olsson, J., Carli, C., Harenstam, 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