NU712 Unit 6 PDSA Cycles Report Example

Three Cycles Before Six Units: A Unit 6 PDSA Report on Testing Two-Way Texts for Patients Newly Home on Insulin

[Student Name]

Purdue University Global

NU712: Quality Improvement and Interprofessional Collaboration for the Doctoral Nurse

Unit 6 PDSA Cycles Report

[Instructor Name]

[Date]

The hospital, units, patients and figures are composite and illustrative; the studies cited are real.

Why Test Before Spreading

The text follow-up was designed in the informatics course to free the hospital's one diabetes educator, whose weekly calls already matched the number of insulin starters leaving the bundle units. Spreading the bundle to six more units would triple that number. The obvious move was to switch on texting for every unit at once. The improvement team chose instead to test it in small cycles, because a tool that looks sound in a design document can break on a patient's phone in ways nobody predicts, and a failure on ten units at once harms more patients than a failure with five.

Taylor et al. (2014) examined 73 published reports of improvement work said to rely on PDSA. Only 14 of them laid out cycles that each built on the one before, and a mere 7 of the 47 examined in full reported collecting numbers monthly or more often to steer each next cycle. Many projects called a single large implementation a PDSA. This report tries to avoid those faults: each cycle had a written prediction, a small scale, data collected during the cycle and a decision that shaped the next.

The Aim and the Measures

The aim of the testing was narrow: by the end of week eight, at least 70 percent of eligible patients leaving 5 West and 6 West on new insulin would enroll, at least 75 percent of those enrolled would reply on day one, and every worrying answer would reach a nurse the same day. The main measures were enrollment, reply rate by day, the number and kind of alerts, minutes from alert to nurse call and nurse minutes spent per week. The balancing measure was nurse time, since a tool that floods the work queue would fail even if patients liked it.

Cycle 1: One Nurse, Five Patients, One Day

Plan. In week one, the 5 West discharge nurse offered enrollment to the next five patients leaving on new insulin. She sent only the day-one message herself from the hospital's secure texting platform, at an hour each patient chose. The team predicted that four of five would reply within six hours and that each patient would cost under ten minutes of nurse time.

Do and study. All five enrolled, but only three replied within six hours. One patient had given his daughter's number and she was at work; the message reached him that evening. A second answered "2, missed a dose" to the insulin question, and the call that followed showed she had taken her dose but read "as your plan says" as asking whether she had followed the meal advice too. Nurse time averaged 11 minutes per patient, mostly on that call and on entering numbers by hand.

Act. The team adapted rather than adopted. The second question was reworded to "Did you take your insulin dose today? 1 yes, 2 no, 3 not sure how much," and the discharge nurse now confirms during teaching whose phone the number belongs to.

Cycle 2: One Unit, Three Weeks, All Three Days

Plan. In weeks two to four, every insulin starter leaving 5 West was offered enrollment, and the platform sent all three messages automatically with the reworded question. The team predicted eight of ten would enroll, three in four would reply on each day and alerts would average no more than one per patient.

Do and study. Ten patients left 5 West on new insulin, and eight enrolled. One had no mobile phone; one preferred a call and was given one. Replies on day one came from seven of the eight, close to the prediction. Replies then fell: six of eight on day three and four of eight on day seven. The platform raised nine alerts across the eight patients. Four came from silence rather than answers. Five were answers that needed a call: two glucose readings under 70, one patient who had run out of pen needles because the pharmacy had filled the wrong size, one who was not sure of her correction dose and one who texted "CALL" because his wife wanted to ask about driving. Median time from alert to nurse call was three hours. Nurse time was 96 minutes over the three weeks, against the roughly two hours the educator's scheduled calls would have needed for the same patients.

The pen-needle problem would not have surfaced until the patient's next visit, and the text found it on day three.

Act. The tool was adopted for 5 West, with two changes for the next cycle: a single reminder two hours after any unanswered message and a test of Spanish messages, since the next unit serves more Spanish-speaking patients.

Cycle 3: Two Units, Four Weeks, Two Languages

Plan. In weeks five to eight, the tool ran on 5 West and 6 West, with reminders and a Spanish version reviewed by two Spanish-speaking patients and the hospital's interpreter service. The team predicted 70 percent enrollment, a day-seven reply rate of at least 70 percent and a Spanish-speaking reply rate no lower than the English one.

Do and study. Twenty-three patients left the two units on new insulin; 18 enrolled, or 78 percent. Fifteen of the 18 replied on day one and 13 on day seven, both rates above the prediction for the first time, which suggests the reminder helped. Four of the 23 patients preferred Spanish. Three of them enrolled and all three replied on every day, a small group that cannot show equal performance but gave no sign of a gap. Eleven alerts fired. Six led to real problems: three lows, one patient who had stopped his basal insulin because his copay was higher than he had been told, one who had the wrong pen needles again and one who could not read her meter's screen.

One failure mattered more than the successes. An alert raised at 1910, after the nurse work queue closed at 1800, was not seen until 0730 the next morning. The patient, who had answered "1, yes" to the question about lows, had treated a reading of 61 correctly and was safe, but the delay broke the rule that every worrying answer reaches a nurse the same day.

Act. Message times were limited so that the six-hour window always closes before 1700, and alerts arriving after 1800 now route to the house supervisor's phone. The repeated pen-needle error was taken to pharmacy, which added needle size to the discharge prescription check.

The Run Chart

The team plotted weekly day-one reply rates for the eight weeks, from 60 percent in week one to between 75 and 88 percent in weeks five to eight. Perla et al. (2011) set out rules for reading run charts; by their rules, a real shift in this chart would take six or more weekly reply rates in a row that all beat the median, or all miss it. Eight weekly points with small numbers in each cannot meet those rules convincingly. The chart is honest about that and is used to show direction, not to claim a proven change. The cycles' value lies in what they revealed and fixed, not in the chart.

What the Cycles Did Not Test

Reed and Card (2016) warn that PDSA is often treated as simple and quick when it is meant to be a disciplined way of learning, and that small cycles are sometimes asked to answer questions they cannot answer. These three cycles did not test whether texting lowers 30-day returns for glucose problems; 38 patients over eight weeks cannot show that. They did not test the tool on Marshallese-speaking or Old Order Amish patients, because none were discharged on new insulin during the cycles. They did not test what happens when the discharge nurse who ran cycle 1 is on vacation. Each gap is now a planned cycle: a Marshallese version in week nine, the community telephone route for Amish families when the next patient is enrolled and a cycle run entirely by night-shift staff.

Decision and Next Cycle

The team's decision is to adopt the text follow-up on 5 West and 6 West and to make it part of the bundle that spreads to the remaining units, carrying the after-hours routing and the needle-size check with it. The next cycle, cycle 4, will run it on one spread unit, 4 West, for four weeks with that unit's own champion, predicting that a unit new to the bundle will reach 70 percent enrollment within a month. Nurse time will be tracked weekly; if it passes three hours a week on any unit, the team will test whether the day-seven message can be dropped for patients whose first two answers were safe.

Conclusion

Three small cycles changed the tool four times before it reached a single new unit: a reworded question, a confirmed phone owner, a reminder and an after-hours route. None of those changes was in the original design, and two of them address safety failures that a hospital-wide launch would have repeated on every unit. The cycles show the tool can be run with modest nurse time and that patients reply when reminded. Whether it reduces returns is a question for the DNP project's evaluation, which these cycles have made worth running.

References

Perla, R. J., Provost, L. P., & Murray, S. K. (2011). The run chart: A simple analytical tool for learning from variation in healthcare processes. BMJ Quality & Safety, 20(1), 46-51. https://doi.org/10.1136/bmjqs.2009.037895

Reed, J. E., & Card, A. J. (2016). The problem with Plan-Do-Study-Act cycles. BMJ Quality & Safety, 25(3), 147-152. https://doi.org/10.1136/bmjqs-2015-005076

Taylor, M. J., McNicholas, C., Nicolay, C., Darzi, A., Bell, D., & Reed, J. E. (2014). Systematic review of the application of the plan-do-study-act method to improve quality in healthcare. BMJ Quality & Safety, 23(4), 290-298. https://doi.org/10.1136/bmjqs-2013-001862