Medical Events, Incident Investigation & Joint Commission Requirements in Radiation Oncology

Incident Investigation, Human Factors & Root Cause Analysis (RCA)

When a clinical error, equipment breakdown, or radiation misadministration occurs, modern patient safety science rejects the obsolete "person approach" (which focuses punitively on individual blame) in favor of the "systems approach" pioneered by James Reason.

James Reason's "Swiss Cheese" Model

Adverse clinical events rarely arise from a single isolated mistake. Instead, complex multi-layered defenses, barriers, and safeguards exist in radiation oncology (e.g., physics plan checks, secondary MU computations, therapist barcode verification, image-guided registration, and physical interlocks). An adverse event occurs only when vulnerabilities—termed holes in the Swiss cheese—align across multiple defensive layers:

   ORGANIZATIONAL          UNSAFE             DEFENSIVE          ACTIVE
     INFLUENCES          SUPERVISION           BARRIERS          FAILURE

   [ Production  ]     [ Inadequate  ]     [ Omitted 2nd ]     [ Incorrect  ]
   [ Pressures / ] ──> [ Staffing /  ] ──> [ MU Check /  ] ──> [ Patient /  ] ──> ADVERSE EVENT
   [ Understaffed]     [ No Protocol ]     [ Bypassed QA ]     [ Overdose   ]
      (Hole)              (Hole)              (Hole)              (Hole)

• Latent Conditions: Systemic, dormant defects embedded within institutional design, software interfaces, organizational staffing, high workload stress, or ambiguous clinical protocols. They remain undetected for months until triggered by operational circumstances.
• Active Failures: Direct unsafe acts, slips, lapses, or rule violations committed by frontline personnel at the sharp end of patient care (e.g., selecting the wrong patient plan on the record-and-verify console).

Root Cause Analysis (RCA) Methodology

Under guidelines from the National Patient Safety Foundation (NPSF) and ASTRO, a formal Root Cause Analysis follows structured steps:

1. Chartering a Multidisciplinary Team: Includes the Qualified Medical Physicist, Radiation Oncologist, Dosimetrist, Radiation Therapist, Oncology Nurse, and Hospital Quality/Risk Director. Personnel directly involved in the event participate as factual witnesses but do not lead the investigative findings to avoid defensive bias.
2. Timeline & Process Mapping: Chronological reconstruction of every clinical action, software timestamp, data transfer, and communication leading to the incident.
3. Ishikawa (Fishbone) Cause-and-Effect Diagram: Systematically categorizes contributing causal factors into distinct domains:

1. Barrier Analysis: Identifies which physical, procedural, administrative, and technological barriers were designed to stop the hazard, why they failed, and what new forcing functions must be installed.
2. Corrective and Preventive Actions (CAPA): Formulates actionable solutions ranked by the Hierarchy of Effectiveness: physical forcing functions (e.g., automated software interlocks blocking delivery without verified secondary checks) are far more effective than administrative rules or staff retraining.

Incident Learning: The ASTRO/AAPM RO-ILS Framework

The Radiation Oncology Incident Learning System (RO-ILS) is a nationwide initiative sponsored by ASTRO and AAPM. Operating under the federal Patient Safety and Quality Improvement Act of 2005, RO-ILS provides legal confidentiality and privilege protection. Clinical practices submit anonymous reports of adverse events and near misses (good catches). Reporting near misses—events that had the potential to cause patient harm but were intercepted by clinical defenses—allows health physicists to identify latent systemic flaws before an actual patient injury occurs.

Regulatory Reporting: NRC Medical Events (10 CFR 35.3045)

Internal incident learning is voluntary and protected; regulatory reporting is legally required. For therapy with byproduct material (HDR and LDR brachytherapy, gamma stereotactic radiosurgery with cobalt-60, and radiopharmaceutical therapy), the NRC or Agreement State defines a reportable medical event in 10 CFR 35.3045. Events that result from patient intervention are reported separately under 35.3045(b) only if they cause unintended permanent functional damage to an organ or physiological system, as determined by a physician.

Criteria Other Than Permanent Implants: 35.3045(a)(1)

(i) Dose difference plus percentage error. The dose differs from the prescribed dose (or the dose that would have resulted from the prescribed dosage) by more than 0.05 Sv (5 rem) effective dose equivalent, 0.5 Sv (50 rem) to an organ or tissue, or 0.5 Sv (50 rem) shallow dose equivalent to the skin, and one of the following is true:

• the total dose delivered differs from the prescribed dose by 20% or more;
• the total dosage delivered differs from the prescribed dosage by 20% or more, or falls outside the prescribed dosage range; or
• the fractionated dose delivered differs from the prescribed dose for a single fraction by 50% or more.

(ii) Wrong administration. A dose exceeding the same thresholds from a wrong radioactive drug or wrong brachytherapy radionuclide, the wrong route of administration, the wrong individual, the wrong mode of treatment, or a leaking sealed source.

(iii) Wrong site. A dose to the skin or an organ or tissue other than the treatment site that exceeds the dose expected under the written directive by both 0.5 Sv (50 rem) or more and 50% or more.

Permanent Implant Brachytherapy: 35.3045(a)(2)

Since the 2018 amendments, permanent implants are judged by source strength rather than dose, and sources implanted in the correct site that later migrate are excluded. A medical event occurs when:

• the total source strength administered differs by 20% or more from the total documented in the post-implantation portion of the written directive;
• the total source strength administered outside the treatment site exceeds 20% of the documented total; or
• the administration involves the wrong radionuclide, the wrong individual, sources implanted directly into a location discontiguous from the treatment site, or a leaking source giving more than 0.5 Sv to an organ or tissue.

Mandatory Notification Timelines

1. NRC Operations Center: telephone notification no later than the next calendar day after discovery (35.3045(c)).
2. Written report: to the appropriate NRC Regional Office within 15 days after discovery. It includes the licensee's name, the prescribing physician's name, a description of the event, why it occurred, the effect on the individual, actions taken or planned to prevent recurrence, and certification that the individual was notified (or why not). It may not contain the individual's name or other identifying information (35.3045(d)).
3. Referring physician and patient: notified no later than 24 hours after discovery, unless the referring physician personally informs the licensee that he or she will inform the individual or that, based on medical judgment, telling the individual would be harmful. The licensee need not notify the individual without first consulting the referring physician, and appropriate medical care may not be delayed (35.3045(e)).

Linacs, State Rules, and The Joint Commission

Medical linear accelerators do not use byproduct material, so their errors fall under state radiation control programs, whose misadministration or medical event rules vary from state to state. Accreditation adds a separate layer. The Joint Commission (TJC) sentinel event policy lists unintended radiotherapy to the wrong body region or more than 25% above the planned radiotherapy dose as a sentinel event. Reporting the event to TJC is encouraged rather than required, but the organization must complete a comprehensive systematic analysis (root cause analysis) and corrective action plan within 45 business days of the event or of becoming aware of it. Diagnostic imaging has a parallel sentinel event for prolonged fluoroscopy with a cumulative peak skin dose above 15 Gy.

Worked Classification Examples

• HDR wrong catheter connection: A $7\text{ Gy}$ HDR fraction is delivered through a transfer tube connected to the wrong channel, giving an estimated $6\text{ Gy}$ to an adjacent bladder region expected to receive $2\text{ Gy}$. The excess is $4\text{ Gy}$ ($4\text{ Sv}$, far above 0.5 Sv) and $200\%$ of the expected dose (above 50%), so it is a wrong-site medical event under (a)(1)(iii).
• Prostate seed placement: Post-implant imaging shows that $25\%$ of the implanted source strength lies in the bladder base because seeds were placed outside the planned volume. More than 20% of the documented source strength is outside the treatment site, so it is a medical event under (a)(2)(ii), regardless of the calculated prostate D90.
• I-131 dosage error: A patient prescribed $3.7\text{ GBq}$ ($100\text{ mCi}$) receives $5.55\text{ GBq}$ ($150\text{ mCi}$) because of a calibrator setting error. The dosage differs by $50\%$, and the resulting extra thyroid-remnant and whole-body doses exceed the 0.5 Sv organ threshold, so it is reportable under (a)(1)(i).

Pitfalls

• Using dose criteria for permanent implants: D90 or other dose metrics are useful clinically but are not the regulatory test for permanent implants.
• Assuming linac errors go to the NRC: linac events are reported under state rules (and handled under TJC policy if the hospital is accredited).
• Including patient identifiers in the written report: the regulation forbids it.