HLTH 5633 Module 3 Measure Set Design Example

Seven Measures for One Aim: Outcome, Process, Structure and the Balancing Measures That Watch for Alert Fatigue in Critical Result Reporting

Student Name

American College of Education

HLTH5633: Advanced Quality Management for Healthcare Administrators

Module 3 Assignment

Instructor Name

July 17, 2028

Why One Measure Is Not Enough

The previous module fixed a precise aim for six medical-surgical units at an invented 260-bed community hospital: nineteen of every twenty critical laboratory values confirmed inside half an hour by the nurse or physician caring for that patient, by the end of June, and almost none, one in a hundred at most, left waiting beyond an hour. The aim's measure is necessary but not sufficient. It says whether results are acknowledged on time, not whether patients receive treatment, whether the new process is actually being used as designed or whether the changes create new problems elsewhere.

Donabedian (1988) proposed that quality can be assessed through structure, the resources and settings of care; process, what is done; and outcome, the effects on patients. Improvement teams have added a fourth category, balancing measures, which look for unintended consequences of a change in other parts of the system (Provost & Murray, 2011). A family of measures covering all four gives a team a way to tell whether a change is working, why and at what cost. The aim measure tells the team whether it is winning; the rest of the family tells it whether the win is real.

Outcome Measures

Outcome one: harm events related to delayed critical results. Numerator: rapid response calls, cardiac arrests or unplanned transfers to intensive care within six hours after a critical laboratory result that had not been acknowledged by the responsible caregiver within 60 minutes. Denominator: all critical results on the six units. Source: rapid response and arrest logs linked to laboratory timestamps. Frequency: monthly, with each event reviewed. Because these events are rare, the measure will be displayed as the number of days between events rather than as a monthly rate, which makes changes visible despite small numbers.

Outcome two: timely treatment of the most dangerous values. Numerator: potassium results above 6.0 mmol/L for which a treatment order, such as calcium, insulin with glucose or a potassium binder, or a documented clinical decision not to treat, occurs within 60 minutes of result finalization. Denominator: all such results. Source: laboratory and order entry timestamps. This measure asks the question the aim measure cannot: whether acknowledgment led to action.

Process and Structure Measures

Process one is the aim measure itself: the share of critical values confirmed by the assigned nurse or physician inside half an hour, reported alongside the share left waiting beyond an hour. Process two: escalation performance. Numerator: calls in which the assigned nurse did not answer within ten minutes and the laboratory escalated to the charge nurse or hospitalist as required. Denominator: all calls in which the assigned nurse did not answer within ten minutes. Source: laboratory call log. This measure shows whether the backstop built into the new process is being used.

Structure one: readiness of the direct-call system. Numerator: shifts on which every nurse on the unit has a working unit phone and the nurse assignment feed to the laboratory is accurate at the start of the shift, verified by a daily spot check of five assignments. Denominator: all shifts. Source: charge nurse checklist. If the assignment feed is wrong, the laboratory will call the wrong nurse, and every process measure will look worse for reasons unrelated to how quickly nurses respond.

Balancing Measures

The changes have two predictable side effects, and each has a balancing measure. The first is alert fatigue. The event analysis found that the hospital had disabled its on-screen critical value warnings some time ago because nurses received too many and began ignoring them. Restoring an alert for the most dangerous values and adding direct calls could recreate that problem. Research on clinical decision support has found that acceptance of alerts declines as clinicians receive more of them and as a larger share are repeats (Ancker et al., 2017). Balancing one: the number of critical result alerts and calls per nurse per twelve-hour shift, from the alert log and call log, with a threshold for review if the median exceeds three per shift on any unit. A short quarterly survey question asking nurses whether critical result notifications interrupt care unnecessarily will supplement the count.

The second side effect is workload in the laboratory. Calling the assigned nurse, waiting up to ten minutes and escalating takes technologists longer than calling a unit desk. Balancing two: median laboratory staff time per critical result notification and the proportion of notifications taking longer than 20 minutes of technologist time, from the call log. If the direct-call process delays other laboratory work, the team will see it here before it appears in turnaround times. A balancing measure is an agreement, made in advance, to look for the harm a change might cause.

Reading the Measures Together

The measures will disagree at times, and the team decided in advance how to read them. If the aim measure improves but timely treatment does not, acknowledgment is becoming a documentation step rather than a clinical one, and the team will review cases with the hospitalists. If the aim measure stalls while the structure measure is low, the problem is the assignment feed or phones, not nurses. If the aim measure improves but alert volume per nurse rises above the threshold, the team will narrow the list of values that trigger an alert before nurses begin to ignore it. The harm outcome will be read slowly, over the full year, because rare events cannot show quick change.

How Much Data Each Measure Costs

A measure family is only sustainable if collecting it does not become a job in itself. Five of the seven measures can be produced automatically from timestamps that already exist in the laboratory system, the call log, the alert log and the electronic record, once an analyst builds the monthly report. The structure measure requires a charge nurse to check five assignments at the start of each shift, about three minutes of work. The harm outcome requires linking rapid response and arrest logs to laboratory results, which the patient safety office already reviews after each event and can extend with one additional field.

The only measure that requires new effort from front-line staff is the quarterly survey question on alert burden, which will be added to an existing unit engagement pulse survey rather than sent separately. Designing the measures around data the hospital already captures is deliberate: the next module addresses how to collect data without asking anyone to work overtime to maintain it, and the choices made here are what make that possible.

Conclusion

One aim, getting critical results to the responsible caregiver within 30 minutes, needs seven measures to be managed well: two outcomes that show whether patients benefit, two process measures including the aim itself and the escalation backstop, one structure measure that checks the system's readiness each shift and two balancing measures that watch for alert fatigue and laboratory workload. Defined precisely and read together, the family tells the team whether the change is working, where it is failing and what it is costing, which a single percentage cannot.

References

Ancker, J. S., Edwards, A., Nosal, S., Hauser, D., Mauer, E., Kaushal, R., & HITEC Investigators. (2017). Effects of workload, work complexity, and repeated alerts on alert fatigue in a clinical decision support system. BMC Medical Informatics and Decision Making, 17, Article 36. https://doi.org/10.1186/s12911-017-0430-8

Donabedian, A. (1988). The quality of care: How can it be assessed? JAMA, 260(12), 1743-1748. https://doi.org/10.1001/jama.1988.03410120089033

Provost, L. P., & Murray, S. K. (2011). The health care data guide: Learning from data for improvement. Jossey-Bass.