Published
- Medical directors
- Patient-safety teams
- Quality leaders
- Physicians
Prepared · sources accessed (UTC) · physician review still to be recorded
A review that ends with “staff reminded” may leave the mechanism of failure unchanged. A review that names only a system problem may overlook an important individual decision. Useful learning requires an accurate chronology, examination of both decisions and conditions, a specific action, and evidence that the action works afterward.
This guide provides a fictional review packet and an action register. It addresses learning after harm, a near miss, or a complaint. Patient care, communication, reporting, professional review, employment action, and litigation preservation may proceed alongside learning, but each has its own purpose and requirements.
Reconstruct what was knowable at each decision
Begin with the event, entry, and availability times of important information. Separate facts known then from facts learned after the outcome. A later diagnosis may explain the event without establishing that the diagnosis was reasonably apparent at an earlier encounter.
Preserve original records through the authorized process. Identify subsequent amendments and the reason for them. Interview participants about the information, constraints, and choices they encountered, while recognizing that recollection can differ from contemporaneous records. Record disagreements and unanswered questions rather than forcing a single confident narrative.
AHRQ’s CANDOR investigation guide offers a system-focused approach to event investigation. Use it as a learning resource; statements about protection within a toolkit do not independently establish legal privilege for a particular review.
Examine individual decisions and working conditions together
Ask what the person did, why the action made sense to them at the time, what alternatives were available, and what information or authority they lacked. Then examine the workflow: task assignment, coverage, interface behavior, training, competing demands, and escalation routes.
An explanation is not automatically an excuse. A system contribution does not erase an individual’s responsibilities, and an individual mistake does not establish that the surrounding design was adequate. Keep the supported findings specific. “The coverage directory contained a former employee’s number” is more actionable than “communication was poor.”
Avoid diagnosing motives from an adverse outcome. If deliberate conduct or a professional concern requires a separate process, refer it through the appropriate procedure with fair handling and accurate evidence.
Include the patient’s account as evidence with a source
Ask what the patient or family was told, what they understood, what they attempted, and what obstacles they encountered. Their account can reveal events absent from institutional records. Identify it as their account and investigate relevant discrepancies rather than dismissing it or treating every detail as adjudicated fact.
Clinical communication after harm should address the patient’s needs and known information through the organization’s appropriate response process. Do not postpone necessary care while an investigation determines every contributing factor. Equally, do not make unsupported statements of causation or promises that the evidence cannot support.
Distinguish a factual update from the final review conclusion. Record what remains under investigation and who will provide the next communication.
Fictional worked packet: an amended report is overlooked
Fictional case. A practice receives an initial imaging report on Monday. A clinician reviews it and communicates a plan. On Tuesday, the imaging service issues an amendment with a recommendation that changes the intended follow-up. The amended report enters the record, but the team’s task list continues to display the earlier item as completed. The patient is later hospitalized; whether the delay contributed to that outcome remains unresolved.
The review first establishes the two report versions and their availability. It asks whether the amendment generated a new task, whether a responsible clinician received it, and whether the patient received updated instructions. Technical staff investigate the configuration; clinical reviewers assess the significance and appropriate response to the amended information.
The patient reports that they called Wednesday because the imaging center mentioned another test. The telephone record confirms a call but summarizes it only as an appointment question. The review retains both sources and investigates what was communicated and routed.
The provisional finding is a gap in identifying and assigning amended reports. It is not yet a conclusion that the gap caused the hospitalization. The corrective proposal is to create a visible amended-result task with accepted clinical ownership and to test that pathway. The causal issue remains with qualified reviewers using the full clinical record.
Convert the finding into a specific action
State the mechanism the action is intended to change. In the example, “retrain everyone” is incomplete because the task list may still conceal the amendment. A more specific action identifies the system configuration, the receiving role, the backup, and the completion evidence.
Name the person who can implement the change and the person who can verify it. They may differ. Set a clinically and operationally appropriate review point. Decide what will happen if the proposed change cannot be implemented or introduces another hazard.
AHRQ’s CANDOR guide emphasizes assigning corrective actions and measuring implementation. The worksheet below turns that general principle into a local record; it is not a validated scoring system.
Action and follow-up worksheet
Use one row for each supported finding. Keep speculative causes in a separate question list until investigated.
| Supported finding and source | Proposed mechanism of improvement | Owner and authority | Implementation evidence | Follow-up measure and denominator | Unintended effect to check |
|---|---|---|---|---|---|
| ______ | ______ | ______ | ______ | ______ | ______ |
| ______ | ______ | ______ | ______ | ______ | ______ |
| ______ | ______ | ______ | ______ | ______ | ______ |
For the fictional case, implementation evidence could include a safe test showing that an amended report creates an assigned task. Follow-up might examine all eligible amendments in a defined period, including those that failed to route. Counting only successful acknowledgments would hide the failure mode.
Reassess after staff turnover, interface changes, and enough routine use to examine whether the improvement persists. Monitor duplicated alerts or unnecessary workload. A successful demonstration shows that the tested pathway can work; it does not guarantee that every future event will be handled correctly.
Keep learning, reporting, discipline, and preservation distinct
Do not promise confidentiality merely because a document is labeled “quality,” “mortality,” or “peer review.” California Evidence Code section 1157, reproduced here, addresses specified committee records and proceedings and contains exceptions. Its application requires attention to the body, document, purpose, proceeding, and governing law.
Federal patient-safety protection is also defined, not automatic. 42 U.S.C. § 299b-21 excludes original medical, billing, discharge, and other original patient or provider records from patient safety work product, and separately maintained information does not acquire that status merely by being reported. HHS’s guidance page addresses these protections and external obligations.
Ask qualified counsel to determine the review structure, access, preservation, reporting, and disclosure requirements. A learning review should not be used to relabel ordinary records or conceal an independent reporting obligation. A Medical Board process, civil claim, employment action, and committee review should not be treated as interchangeable findings or procedures.
Questions readers ask
Does a system-focused review mean no one is accountable?
No. It examines individual actions and the conditions that shaped them. Accountability should follow supported facts and the appropriate process, while improvement addresses the mechanisms identified.
Is a near miss worth reviewing if no harm occurred?
Yes, when it reveals a consequential vulnerability. The review can examine why the event was intercepted and whether that protection is reliable. Do not invent a harm outcome to justify learning.
Does the absence of another reported event prove the fix worked?
No. Reporting may be incomplete and the event may be infrequent. Check whether the changed process was implemented and examine a defined set of relevant tasks, including failures and missing information.
Related resources and sources
Existing foundations: clinical governance, policy and law, and physician due process. Continue with workflow responsibility, measurement methods, and the separate peer-review and mortality-review guide.
Source and scope note. Sources include AHRQ CANDOR, the official federal statutory compilation, HHS guidance, and a reproduction of California section 1157 because the official endpoint was unavailable. This is a learning method, not a privilege opinion or a determination of liability. Prepared November 5, 2026 Pacific.