Appendix F
Appendix F — Benefit–Burden–Safety Measurement Plan
Decide what the findings can change
Use with: Chapters 10, 14, and 15, and Reference Chapters R1 and R8.
Choose measures that help decide whether to continue, adapt, stop, or expand a service. Include benefit, burden, safety, access, a baseline, responsibility for review, and the people’s own accounts.
F1. The measurement plan
Field | What to define before the review |
|---|---|
Human purpose | What matters to the people receiving or providing the support, and which question the evaluation should answer. |
Population and setting | Who is included, who may be missed, and which access routes are available. |
What is being tried | The actual service, content, staffing, and workflow version—not only its name. |
Baseline and comparison | What happens now and what comparison can support the intended conclusion. Record missing baseline information honestly. |
Benefit | A specific result to examine: understanding, received support, useful participation, caregiver relief, or another relevant outcome. |
Process | What action should occur and how it will be confirmed. Do not substitute this measure for the intended benefit. |
Burden and resources | Family effort, travel, cost, staff preparation, review, correction, duplicate entry, and after-hours work. |
Safety and privacy | Errors, missed responses, inappropriate pressure or sharing, and the responsible route for addressing concerns. |
Equity and access | Who could not use the offer, who withdrew, and what is known about the reason. Do not infer demographics or preferences. |
Measure specification | Unit, numerator and denominator where relevant, data source, instrument/version, language, respondent, and handling of missing information. |
Period and review | Enrollment and follow-up periods, frequency, named reviewer, and the decision the result can change. |
Action criteria | What prompts investigation, correction, pause, or expansion; who has authority; and how ongoing care is protected. |
Use a few measures well. Do not require every family to complete a new survey when an authorized existing record can answer the operational question. When a validated instrument is appropriate, use its required wording and scoring rather than an AI rewrite.
F2. Keep different results separate
Distinguish reach, understanding, a first chosen action, adoption, useful participation, outcomes, and attributable impact. Also distinguish a willingness to recommend, an actual referral, and another person receiving help. The behavioral review uses these distinctions to prevent activity counts from becoming unsupported benefit claims.
Hypothetical arithmetic example: Twelve requested conversations were due during a review period. Eight occurred, two were still awaiting a response, one person declined the conversation, and one outcome remained unknown. Report eight of twelve completed conversations, with the other categories visible. The two unanswered requests still need an owner. This count says nothing by itself about caregiver strain or health improvement.
Use the nine original outcome areas as questions to investigate: culture of care, integration of social needs, care equity, community connections, relationships, quality of life, reduced burnout, improved care outcomes, and organizational sustainability. They are not nine guaranteed effects.
The Jack Scale remains a historical and proposed review tool, not an established clinical assessment. These appendices do not use it to rank families, staff, or entitlement to help.
F3. Decision and remeasurement record
Record the finding and date; evidence and uncertainty; the contributor’s account; the decision; the named decision-maker; the change actually made; its date; and the later result. Log reasons for no change as well as changes made. Keep unfavorable and burdensome results.
A report that ends with “the team will review” needs a named person and a review point. This follows the dated improvement-register specification, which also distinguishes that register from the product itself.
F4. Shared Learning Brief
Prepare a version another organization can understand without unnecessary private records:
Our question and setting: [need, population, resources, and access conditions].
What we offered and changed: [actual intervention and versions].
How we examined it: [baseline, comparison, measures, period, missing information].
What happened: [favorable, unchanged, adverse, and burdensome findings].
What we cannot conclude: [uncertainty and other plausible explanations].
What happens next: [decision, owner, next review].
Sources and contributors: [credit, permissions, funding interests, and approved sharing].
Use common definitions before comparing organizations. A service shortage deserves a resource decision, not a ranking of the families unable to obtain care.
F5. Caregiver assessment and reassessment card
Complete this card before collecting sensitive assessment data. It is a local planning tool, not a new questionnaire or a requirement to administer every instrument named in R1.
Planning field | What must be specified |
|---|---|
Purpose and measure | The caregiver’s need; tool, version, language, permissions, population fit, and the professional interpreting it. |
Starting point | What baseline means here: enrollment, first assessment, or another defined event. Record the actual date. |
Return points | Dates and their anchor event. RUSH’s published example includes one and three months after completion or other agreed times (CG1–CG2); do not substitute a different schedule silently. |
Response and safety | Named reviewer, review window, agreed assistance, and appropriate urgent escalation. Do not postpone a needed response until the next assessment. |
Choice and missingness | A route to decline a nonmandatory question or ask privately; distinguish declined, not asked, missing, and answered. Follow instrument rules; do not impute zero need. |
Privacy and access | Where answers are stored, who can access them, what may enter a shared plan, and the tested exclusions from AI inputs and family outputs. |
Evaluation and ownership | Denominator, follow-up coverage, burden, access, adverse findings, and the person authorized to change the process. Local pre/post change is not proof of cause. |
F6. Pathway flow and unresolved demand
Use these optional local measures for a constraint-focused project. Keep the caregiver outcomes and privacy arrangements in F5. The definitions below are proposed planning conventions, not a validated questionnaire or a TOC certification. IHI’s outcome, process, and balancing-measure approach provides the wider evaluation discipline.75
Measure | Definition to agree | Protection against a misleading result |
|---|---|---|
Time to substantive response | Elapsed time from a defined request event to a response addressing the need or agreeing its next handling. | An automatic receipt is not necessarily substantive. Report the starting date, working/calendar time, and unanswered requests. |
Time to confirmation; time to service | Measure separately from the same defined starting event to an accepted arrangement and to actual delivery. | Do not rename an intake date as a referral date. Specify the endpoint and include missing dates. |
Service delivery rate | Defined appropriate services actually delivered per week or another stated period. | Keep people, visits, and service hours distinct. This is not financial throughput or a measure of care quality. |
Unresolved demand | Open requests at the review date, their age, stage, known reason, and responsible responder. | Do not erase a waiting person by closing a record, transferring it to another queue, or calling no response a decline. |
Handoff reliability | Due handoffs reaching a specified stage divided by all handoffs due for that stage in the review period. | Recipient acceptance, confirmation, and service receipt are separate stages; retain unknown and overdue outcomes. |
Total effort and human result | Staff and family coordination, setup, review, correction, travel, cost, and the relevant experience or outcome. | Pair process improvement with safety, privacy, access, appropriate priority, and the person’s chosen goals. |
Hypothetical arithmetic example. Of 12 requests due for a substantive response, eight received one, two are waiting, one person declined further contact, and one outcome is unknown. The response proportion is 8/12. Five visits delivered that week means a delivery rate of five visits—not eight responses or 8/12 visits. These counts alone establish no clinical or caregiver benefit.
Report completed-case waiting times alongside the ages of open requests; show longer waits when the sample supports that description. Define missing, declined, not applicable, and still waiting separately. Preserve clinically appropriate priorities. Investigate whether faster service came from shifting work, narrowing access, or shortening a conversation that people needed.
Verify data availability before promising a measure. Missing referral or completion dates must remain missing. Appendix J records the dated field limitations; check later versions rather than invent dates or assume measurement is automated.
Retain F1’s measure definitions, comparison, period, source, effort, reviewer, and action threshold. Examine adverse, unequal, and burdensome findings as well as benefits. A pre/post pattern informs learning without proving cause. Decide whether the support is useful, safe, accessible, and sustainable enough to continue; see R1’s measurement guidance.
Notes
Institute for Healthcare Improvement. Model for Improvement: Establishing Measures. Guidance on outcome, process, and balancing measures. Source (opens a new tab)