HI501 Unit 5 Usability Walkthrough Example

A Bedtime Glucose of 247 and a Scale on the Wrong Tab: A Unit 5 Usability Walkthrough of Correctional Insulin Documentation, Rated Against Nielsen's Heuristics

[Student Name]

Purdue University Global

HI501: Health Informatics

Unit 5 Usability Walkthrough

[Instructor Name]

[Date]

The ward, the nurses and the screens are composite; the methods and sources cited are real.

Introduction

Correctional insulin, given before meals and at bedtime when blood glucose is above target, is one of the most frequent medication tasks on a medical ward. It is time-sensitive, it depends on a number the nurse must look up, and it is easy to get wrong by one row of a scale. That combination makes it a good task for a usability walkthrough: small enough to bound, common enough to matter and risky enough to justify attention.

Method

The task was bounded tightly: from a point-of-care glucose result of 247 at 21:30 to a documented correctional insulin dose. I performed a cognitive walkthrough in the training environment, answering at each step the four questions the method asks: will the user try to achieve the right effect, will the user notice the correct action is available, will the user associate the action with the effect, and will the user see progress after acting (Polson et al., 1992). I then observed two nurses completing the same task on a live unit at bedtime, recording steps and seconds, with no patient identifiers collected. Problems were rated against Nielsen's ten usability heuristics using his five-point severity scale, from 0 for no problem to 4 for a usability catastrophe (Nielsen, 1993). One walkthrough and two observed nurses are enough to find problems and judge their seriousness. They are nowhere near enough to estimate how often errors occur, and this paper makes no such claim.

The Step Sequence

1. Open the patient's chart from the unit list. Screen 1.

2. Open the medication administration record. Screen 2.

3. Find the correctional insulin line among the scheduled medications.

4. Select it to administer, which opens the dose window. Screen 3.

5. Scan the patient's wristband.

6. Scan the insulin pen.

7. Read the dose window, which shows "Dose: per scale" and an empty units field.

8. Look for the glucose result, which the dose window does not show; open the results tab. Screen 4.

9. Return to the dose window, which now asks for units.

10. Look for the scale, which is not in the dose window; open the order details tab. Screen 5.

11. Find the bedtime scale, read the row for 201 to 250, and remember the number.

12. Return to the dose window, rescan the pen after a timeout, and type the units.

13. Sign the administration.

In the walkthrough, the task took 13 steps across 5 screens and about 96 seconds. At steps 8 and 10, the answer to the walkthrough's second question was no: nothing on the dose window tells the nurse where the glucose or the scale can be found. The two numbers the nurse needs to choose a dose, the glucose and the scale, are both missing from the one screen where the dose is entered.

Findings

1. Recognition rather than recall, step 10, severity 3. The dose window asks for units while the scale sits on another tab. The nurse must carry the number in memory across a screen change. One observed nurse avoided the problem by writing the scale on a paper towel at the start of the shift, a workaround that will go out of date if the order changes.

2. Error prevention, step 12, severity 3. The units field accepts any number, including decimals such as 2.5, although insulin pens deliver whole units, and it does not check the value against the scale row for the documented glucose.

3. Consistency and standards, step 11, severity 3. The mealtime and bedtime scales are both titled "Correctional insulin" and differ only by a small "HS" at the end of the bedtime line. At 247, the bedtime scale calls for 2 units and the mealtime scale for 4. The second observed nurse first read the mealtime row, then corrected it after a second look.

4. Visibility of system status, step 12, severity 2. The dose window timed out after 60 seconds while the nurse was on the order details tab, cleared the pen scan without saying so, and required a rescan.

5. Flexibility and efficiency, step 8, severity 2. The glucose result, already in the record from the connected meter, is not shown in the dose window, so every nurse makes the same detour.

Usability problems in clinical software are safety problems as well as efficiency problems, which is why heuristic methods developed for general software have been adapted to medical devices and patient safety (Zhang et al., 2003). Three of these five findings could produce a wrong dose.

What the Two Observed Nurses Did

The live observations added what a walkthrough cannot: the strategies people invent to cope. The first nurse, with eleven years on the unit, opened the order details tab before scanning anything, read the scale, then went to the dose window, so the timeout never caught her. She described this as "doing it backwards so the system cannot rush me." The second nurse, three months out of orientation, followed the screens in the order they appear and was caught by every problem in turn: the detour to results, the timeout and the near miss with the mealtime row. Her task took 141 seconds. Experience did not make the design safer; it taught one nurse to work around it, and the workaround depends on a habit that new staff do not yet have.

Recommendation

One change addresses findings 1, 2, 3 and 5 together. The dose window should show the most recent glucose with its time, the name of the active scale in full words, and the scale row that matches that glucose, with the corresponding dose prefilled and editable. The units field should accept whole numbers only and ask for a reason when the entered dose differs from the scale row. With those changes, the task would take about 8 steps on 2 screens, and the nurse would never need to remember a number across screens. The timeout in finding 4 should be lengthened for the dose window, or the scan should survive a tab change.

Limits and Next Steps

This walkthrough studied one task in one record design with one evaluator and two nurses. A heuristic review by two or three additional evaluators would find problems this one missed, and a test of the redesigned window with five to eight nurses would show whether it works as intended. A measure for after the change would be the share of correctional doses that match the scale row for the documented glucose, taken from administration records. The finding most likely to change practice, though, is already clear from this small study: the dose window asks the nurse for an answer while hiding both of the facts the answer depends on.

References

Nielsen, J. (1993). Usability engineering. Academic Press.

Polson, P. G., Lewis, C., Rieman, J., & Wharton, C. (1992). Cognitive walkthroughs: A method for theory-based evaluation of user interfaces. International Journal of Man-Machine Studies, 36(5), 741-773. https://doi.org/10.1016/0020-7373(92)90039-N

Zhang, J., Johnson, T. R., Patel, V. L., Paige, D. L., & Kubose, T. (2003). Using usability heuristics to evaluate patient safety of medical devices. Journal of Biomedical Informatics, 36(1-2), 23-30. https://doi.org/10.1016/S1532-0464(03)00060-1