NSG/541 Week 5: Analysis and Interpretation, sample paper

A Six-Minute Wait at Shift Change: Analyzing 18,019 Call Events and Eleven Falls, and What the Numbers Can and Cannot Say

[Student Name]

University of Phoenix

NSG/541: Data Analysis and Management

Week 5 Assignment

[Instructor Name]

[Date]

The hospital, the unit and all data are a composite written for a model paper.

The prepared data set now holds 18,019 call events from 90 days, linked to patients, to eleven falls and to shift staffing. This paper reports the analysis.

Measure 1: Bedside Response Time

Across 15,497 bedside-canceled call events, the median response time was 2.9 minutes and the 90th percentile 9.4 minutes. The median met the committee's target of under 3 minutes, but one patient in ten waited more than nine minutes. Response time varied sharply by hour. From 0600 to 0800 and from 1800 to 2000, the medians were 6.1 and 5.4 minutes, with 90th percentiles over 17 minutes. From 2300 to 0500, the median was 1.8 minutes. The unit's average looks acceptable because its quiet hours are fast; its slowest hours coincide with shift change.

Measure 2: Prompt Response to Bathroom Calls

Of 3,406 bathroom call events, 61% were answered at the bedside within 3 minutes, against a target of 80%. At shift change hours, only 38% were. Desk-canceled bathroom calls, which count as not promptly answered under the Week 3 definition, made up 11% of bathroom calls.

Measure 3: Desk Cancellation Rate

Desk cancellations were 14.0% of all patient calls, against a target of under 5%. They were most common on evening shift, 19%, and least common overnight, 8%.

Measure 4: Fall Rate

Eleven falls over 2,591 patient days give a rate of 4.2 falls per 1,000 patient days, with two injurious falls, a rate of 0.8 per 1,000 patient days. Because the numbers are small, the rate for any single month varied widely, from 2.3 to 5.8.

Linking Calls to Falls

Of the eleven falls, five, or 45%, were preceded within 20 minutes by a call from the patient's bed or bathroom that had been open more than 5 minutes. Four of those five were elimination-related falls, and three occurred between 0600 and 0800 or 1800 and 2000. Of the six falls not preceded by a waiting call, four involved patients who did not call at all.

Workload and Response Time

To see whether waits reflected staffing, I compared median response time with the number of calls per nurse per hour and with nurses on duty. Median response time rose as calls per nurse rose, and it was higher at shift change than at other hours with similar call volume per nurse. Staffing levels, which were similar across day shifts, did not explain the variation within the day. This pattern fits earlier work showing that the pace of activity on a unit, how often patients call and how quickly they turn over, shapes waits independently of nursing hours (Tzeng & Larson, 2011).

Interpreting the Link Between Waits and Falls

The finding that five of eleven falls followed a long wait is consistent with the committee's concern and with published evidence associating faster response with lower fall rates across units (Tzeng et al., 2012). But the data cannot show that the waits caused the falls. Patients who call often may be more at risk for other reasons; the eleven falls are too few to rule out chance; and four falls involved patients who never called, which points to a different problem, patients who try to get up without asking. The honest reading is that long waits, especially for the bathroom at shift change, are a plausible and fixable contributor to some falls, and that other falls need other responses.

Comparing With Published Patterns

The unit's falls resemble those described in a prospective hospital study in which half of falls were elimination related, most occurred in patients' rooms and more than half happened in the evening or overnight (Hitcho et al., 2004). On our unit, six of eleven falls were elimination related and seven occurred in patients' rooms, but more occurred around shift change than overnight. That difference matters: it suggests that our unit's risk window is the handoff period rather than the night, which points to a specific, local solution.

Distribution of Waits

The distribution of bedside response times was strongly right-skewed. Half of calls were answered within three minutes and three-quarters within five, but the longest tenth stretched from about nine minutes to the sixty-minute cap. That tail is where risk concentrates, which is why the 90th percentile is reported beside the median and why a single average would mislead.

A Display for the Committee

The committee will see three displays: a bar chart of median and 90th percentile response time by hour of day, showing the two shift change peaks; a run chart of the monthly fall rate with its median, showing that month-to-month variation is within what small numbers produce; and a simple table of the eleven falls with activity, whether a call was waiting and the hour. Each display answers one question, and each is labeled with the number of events it rests on.

Response by Call Type

Bathroom calls had a median bedside response of 3.6 minutes, standard bed calls 2.7 minutes. Bathroom calls are a smaller share of volume but a larger share of long waits: they were 19% of call events and 27% of calls that waited longer than ten minutes. That imbalance supports focusing improvement on bathroom calls first, since they are both the calls most tied to falls and the calls most likely to wait.

Response by Shift

Night shift had the fastest median response, 2.1 minutes, day shift 3.1 minutes and evening shift 3.4 minutes. Evening shift also had the highest desk cancellation rate, which may make its bedside response time look better than patients' experience, since some of its slowest responses were canceled from the station and excluded from the bedside measure.

Sensitivity Checks

Two checks tested whether data decisions changed the conclusions. Recalculating Measure 1 without merging duplicates changed the median by only 0.1 minute. Using a 30-minute rather than 20-minute window for Measure 5 added one more fall to the "preceded by a waiting call" group, without changing the pattern. The conclusions are stable under these choices.

What the Analysis Did Not Find

The analysis found no difference in response time between rooms near and far from the nurse station, which the committee had suspected, and no weekday-weekend difference. Reporting findings that did not appear is as important as reporting those that did, since it prevents solutions aimed at the wrong problem.

Conclusion

The unit's typical response time meets its target, but bathroom calls are answered promptly only 61% of the time, desk cancellations are common and the longest waits cluster at shift change. Five of eleven falls followed a long wait, four of them elimination-related, while four falls involved patients who did not call. The data support targeting bathroom calls at shift change and desk cancellations, and they do not support claims that response time alone explains the unit's falls. Week 6 will turn these findings into recommendations.

References

Hitcho, E. B., Krauss, M. J., Birge, S., Dunagan, W. C., Fischer, I., Johnson, S., Nast, P. A., Costantinou, E., & Fraser, V. J. (2004). Characteristics and circumstances of falls in a hospital setting: A prospective analysis. Journal of General Internal Medicine, 19(7), 732-739. https://doi.org/10.1111/j.1525-1497.2004.30387.x

Tzeng, H.-M., & Larson, J. L. (2011). Exploring the relationship between patient call-light use rate and nurse call-light response time in acute care settings. CIN: Computers, Informatics, Nursing, 29(3), 138-143. https://doi.org/10.1097/NCN.0b013e3181fc41d9

Tzeng, H.-M., Titler, M. G., Ronis, D. L., & Yin, C.-Y. (2012). The contribution of staff call light response time to fall and injurious fall rates: An exploratory study in four US hospitals using archived hospital data. BMC Health Services Research, 12, Article 84. https://doi.org/10.1186/1472-6963-12-84