In 2005, the Joint Commission established a new reviewable sentinel event tied to radiation exposure.1 This event targets both radiotherapy and fluoroscopy doses that can result in serious patient injury. The fluoroscopic component addresses prolonged fluoroscopy procedures that result in peak skin doses greater than 15 Gy. This new event requires institutions to identify and investigate patient radiation exposures that potentially meet this criterion. Currently, fluoroscopic equipment offers no direct measure of peak skin dose; thus, an investigation is required to collect the necessary information for a dose calculation. This calculation is neither trivial nor straightforward and can require significant effort by the medical physicist. The purpose of this article is to provide a framework for detecting potential fluoroscopic sentinel events and calculating the associated patient skin dose.
The Joint Commission considers a sentinel event to be an unexpected occurrence involving the risk or actuality of death or serious injury to a patient.1 The Joint Commission employs sentinel events to signal the need for immediate investigation and response by the accredited health organization. Sentinel events and medical errors are not the same; a sentinel event may not be the result of an error, and an error may not cause a sentinel event.
Although each hospital is expected to identify and define sentinel events pertinent to its particular environment, the Joint Commission specifically defines 10 types of events to be reviewable. Examples include incidents in which patients suffer abduction, rape, suicide, surgery on the wrong body part, or incompatible blood transfusions. The fluoroscopic event discussed here is the newest reviewable event.
In 2005, the Joint Commission added the following reviewable sentinel event:1 “Prolonged fluoroscopy with cumulative dose >1500 rads to a single field or any delivery of radiotherapy to the wrong body region or >25% above the planned radiotherapy dose.”
The 15 Gy (i.e., 1500 rad) fluoroscopic aspect of this sentinel event is intended to address skin injuries,2 such as serious burns or necrosis resulting from high levels of radiation delivered during fluoroscopy procedures.
The term “cumulative dose” used by the Joint Commission in reference to fluoroscopy differs from the standard usage in clinical medical physics. Technically, cumulative dose refers to the integrated reference-point air-kerma that must be displayed by modern fluoroscopy devices during a procedure.3 In reference to the fluoroscopic sentinel event (FSE), the Joint Commission has provided guidance for their meaning of cumulative dose in an on-line set of frequently asked questions.2 Cumulative dose as used by the Joint Commission is the peak radiation dose delivered to the patient’s skin. Furthermore, it is made clear that cumulative dose to the Joint Commission refers neither to a single procedure nor to a lifetime acquired dose, but that "monitoring cumulative dose over a period of six months to a year would be reasonable." The selection of the specific monitoring time window is left up to the institution.
There are practical consequences of the Joint Commission definition of cumulative dose. Direct measurements of skin dose are not normally available. Therefore, a medical physicist must estimate the cumulative peak skin dose (PSD) using exam data, such as operating parameters, procedure notes, and recorded images. These calculations must assess the highest dose delivered to any part of the patient’s skin over all the fluoroscopic procedures performed within the institution’s selected time window. Monitoring dose over a time window raises questions on how to deal with multiple procedures within a single institution or across multiple institutions. Obtaining information from outside institutions is very problematic, particularly if several months have elapsed. Solution of this problem awaits widespread implementation of portable electronic records.
Radiation injuries to living organisms are related to dose and can be divided into two broad classes: deterministic and stochastic.4 The fluoroscopic sentinel event addresses deterministic injuries, in particular cutaneous radiation injury (CRI). Deterministic radiation injuries occur above a threshold dose level, and the severity of the injury worsens as dose is increased above that level. CRI effects such as erythema and epilation can be observed when x-ray skin doses exceed 2 Gy, and severe wounds can be expected for dose levels above 10 Gy.5 However, there is wide variability in the occurrence, progression, and severity of CRI from patient to patient.
Most diagnostic fluoroscopy procedures are of short duration, and the skin doses received by patients are well below CRI threshold levels. However, fluoroscopically-guided interventional (FGI) procedures may require the prolonged use of fluoroscopy. Complex FGI procedures can result in PSD levels high enough to cause CRI.6
When a sentinel event has occurred, the Joint Commission requires an institution to conduct a timely and thorough root cause analysis (RCA). The RCA must occur within 45 days of the event or the date at which the institution became aware of the event. An RCA focuses on hospital processes, not on assigning blame or assessing individual performance. The aim of an RCA is to identify strategies to prevent similar sentinel events from occurring in the future. The institution must implement and monitor the effectiveness of improvements specified during the RCA.
The Joint Commission encourages, but does not require, that an institution notify the Joint Commission when a sentinel event occurs and an RCA has been held. If the institution does not notify the Joint Commission, then it must be prepared to provide the RCA report at any time the Commission requests it (e.g., at the next Joint Commission inspection). Failure to pursue an adequate and timely RCA and produce an action plan could result in loss of accreditation.
Direct and comprehensive skin dose measurements are rarely available real-time in the clinical setting. Therefore, the cumulative PSD of interest to the Joint Commission is not available to physicians or technologists at the completion of a procedure. Currently, PSD evaluations require investigation and calculation by a medical physicist on a case-by-case basis, and they cannot be performed for every procedure. To efficiently identify potential FSEs requiring detailed evaluation, it is necessary to establish monitoring procedures using PSD indicators that the technical staff can directly access or record at the time of the procedure.
The most common dose indicators found on fluoroscopic equipment are reference point air-kerma (Ka,r), air-kerma area product (PKA), fluoroscopy time, and number of angiographic frames.6 As illustrated in Fig. 1, Ka,r and PKA are measures of radiation output. Ka,r is air-kerma at a reference point along the beam axis of a fluoroscope. PKA is a less direct dose indicator, since it must be divided by beam area to yield air-kerma at a position on the beam axis. Fluoroscopy times and angiographic frame counts are the least useful dose indicators, since they do not account for variations in output radiation that occur with different operating factors and patient size.7,8 On older machines, fluoroscopy time may be the only available indicator.
Displays of the Ka,r delivery rate and the cumulative Ka,r value have been an FDA requirement on fluoroscopy equipment since 2006. Ka,r reference points are defined depending on the fluoroscope type and are intended to reflect the incident air-kerma at the patient under typical operating conditions. Fluoroscopy systems designed for FGI procedures feature isocentric c-arm gantries. This design facilitates changing view angles without having to adjust table position for anatomy positioned at isocenter. As illustrated in Fig. 2, the FDA recommended reference point on isocentric c-arm systems is 15 cm toward the focal spot from isocenter,3 a position commonly known as the interventional reference point (IRP).9
It is important to note that cumulative Ka,r is not the patient PSD. Ka,r does not account for the actual position of the patient’s skin with respect to the IRP, for dose spreading on the patient due to changes in geometry (i.e., view angle and table position), for attenuation by patient support fixtures, or for correction factors required to convert air-kerma to skin tissue dose. Nevertheless, of the commonly available dose indicators, cumulative Ka,r exhibits the best correlation with PSD,10 and the NCRP recommends using it for monitoring deterministic CRI effects.6
Each institution must select threshold levels for dose indicators at which investigations will be triggered. The number of threshold indicators will depend on the quality of dose information available from the fluoroscopy systems in use. If cumulative Ka,r is available, it should be used as the investigation trigger; if older fluoroscopy systems are used, a fluoroscopy time trigger may also be necessary. Regardless, trigger thresholds should be low enough to catch all real events but high enough to avoid unnecessary physics evaluations.
Published data from the RAD-IR study of interventional procedures can be used to establish trigger levels.11,12 In Part 2 of the study, PSDs estimated with a commercial software application were compared to cumulative Ka,r, and a mean ratio of PSD to cumulative Ka,r of about 0.4 was found. However, ratios up to 2.8 were observed for some types of procedures, suggesting that a threshold value of about 6000 mGy for cumulative Ka,r could capture all potential FSEs.12 Alternatively, NRCP Report 168 recommends the use of substantial radiation dose levels (SRDL) for the purposes of overall procedure dose monitoring and post-procedure patient management.6 SRDL values for Ka,r, PKA, and fluoroscopy time of 5000 mGy, 500 Gy-cm2, and 60 minutes, respectively, are given as levels at which substantial risk of cutaneous injury may exist. If trigger levels are required for PKA or fluoroscopy time, ratios between the SRDL values can be used to calculate thresholds consistent with the chosen cumulative Ka,r. Site experience and additional sources of published literature may allow refinement of the initial values, depending on the nature of the practice and procedure types performed.13
Special considerations may be required on biplane fluoroscopes to account for possible overlap of the frontal and lateral radiation fields. A simple, conservative remedy for the purpose of triggering a dose investigation is to add dose indicator values from each plane for comparison with the corresponding threshold.
Tracking of multiple procedures over time requires that a cumulative record of patient dose indicators be maintained. Most current radiology information systems do not yet offer this capability, so a custom database solution may be necessary. Additionally, not all fluoroscopic procedures have a significant potential for large skin doses and may not need to be tracked. A practical approach is to track cases in areas that have significant potential for large skin doses, such as FGI procedures performed in neuro-angio, cardiac catheterization, and electrophysiology labs.
An overall policy for detecting and investigating fluoroscopic sentinel events should be developed and implemented by each institution. One possible scheme is shown in Fig. 3. If information on prior exposures is available from the tracking database, the technologists or nursing staff should notify the physician of previous fluoroscopy time or cumulative Ka,r for the patient before beginning the procedure. Once the fluoroscopic procedure has started, the physician should be appraised of dose indicator levels during the case.6 The physician may then choose to implement extra dose control actions, such as using lower dose modes and changing c-arm angle. In teaching hospitals, the attending may need to take direct control of the procedure if the threshold is approached.
After the procedure, the dose tracking database is updated and the cumulative record of dose indicators is evaluated to determine if an investigative threshold were passed. If a threshold limit is exceeded, the Radiation Safety Officer (RSO) should be notified and an entry made in the hospital event recording system. A defined staff member should be made responsible for all these notifications. All original data for the case should be write-protected on the modality console and preserved until the physics investigation is complete. In the example policy, the RSO notifies the medical physicist that a dose evaluation is required. The results of the investigation are reported back to hospital administration through a defined path. If the physicist determines that the Joint Commission limit on cumulative PSD (15 Gy) has been exceeded, an RCA is held.
In an FSE investigation, the organ of interest is the skin, and the dose quantity of interest is the PSD. PSD can be estimated from dose indicators using published regression formulas for similar FGI procedures.10 However, this method is not patient specific and does not account for circumstances which may contribute to higher than expected doses. A patient specific PSD should be estimated by a medical physicist using information for each procedure involved in the investigation.
Fig. 4 shows the general steps involved in producing a patient-specific PSD for an FSE investigation. A dose delivery timeline detailing c-arm orientation, table positioning, and operating factors is assembled to allow a skin dose map for each procedure to be constructed. A cumulative skin dose map is made by summing the dose distributions from all procedures. Currently, the primary recorded information available for producing skin dose maps is limited to cumulative air kerma indicators and a set of recorded acquisitions (e.g., DA, DSA, or rotational “runs”) with their associated technical factors available in DICOM format. The challenge for the physicist is to reconstruct skin dose maps from this limited information.
The information required for a PSD calculation is usually incomplete and must be collected from different sources (Fig. 5). The best source of electronically recorded information is most often the modality device, where dose indicator information, logged notes, original images, and other critical information can be accessed. Since storage capacity is usually limited on modality devices, staff should be advised to protect any information associated with an investigation.
Currently, the most important information for a PSD calculation in conjunction with Ka,r is the set of acquired images in DICOM format.14 On interventional systems, angiographic acquisitions (e.g., DA, DSA, or rotational “runs”) are automatically saved.6 Modern systems also provide the ability to store fluoroscopy images; however, storage space restrictions limit the number of images that can be saved. At a minimum, saved images will show patient position, anatomy imaged, and field size. In addition, DICOM image files contain useful dose and geometry related information (Fig. 6). The best source for DICOM images is on the modality itself, since not all acquisitions may be sent to a PACS for long-term storage and some PACS exclude or modify DICOM information. However, the PACS should always be checked for previous procedures. The DICOM standard also supports structured dose reports.15 If available, these reports are another valuable source of information for fluoroscopic and angiographic doses.
A physical inspection of the fluoroscopic equipment should be performed. The inspection should verify the Ka,r calibrations, associated reference location, and other pertinent operating parameters. Phantom-based dosimetry may be necessary to construct a dose-delivery timeline when recorded technical factors are inadequate. The physicist should also observe similar procedures to gain knowledge of operating practices and patient positioning techniques.
Staff interviews are important sources of information and should be conducted as soon as possible. Staff can provide information not available in recorded form, such as the timeline, anatomy imaged during fluoroscopy, and details of patient positioning. A large amount of patient information may be recorded in the hospital and radiology information systems (i.e., HIS and RIS). In the patient records, diagnostic reports for the procedure and patient biometrics relevant to a PSD estimate are available. Patient images from other modalities, such as CT, may provide additional body habitus information.
As shown in Fig. 7, the methodology for reconstructing a dose delivery timeline at a reference position varies depending upon the available information. Systems with a dose meter and saved run images provide the most information and require the least number of calculations and assumptions. In the best case scenario, the DICOM dose information associated with a run includes the preceding fluoroscopy dose, and the timeline is directly available. In cases where run dose does not include fluoroscopy, the total fluoroscopy dose can be calculated from the difference of the cumulative dose and the summed run doses. If the run DICOM dose is not available, it can be calculated based on DICOM technique factors, number of frames, and tube output measurements. In the worst case scenario, fluoroscopy dose must be estimated based on fluoroscopy time and phantom measurements using typical dose modes after the run doses have been calculated.
Regardless of the information available, assumptions must be made to assign fluoroscopy dose to the timeline. In all cases, an inherent assumption is that fluoroscopy was performed at the same c-arm angles and beam sizing as angiographic runs. Absent any specific timing information of fluoroscopy dose, additional assumptions to apportion the cumulative fluoroscopy dose along the timeline are required. A reasonable approach could divide the fluoroscopy dose between runs with weighting determined by the dose-rate for each associated run. The weighting scheme could also be refined to limit fluoroscopy dose between runs to that possible given the calculated fluoroscopy-rate and the time between runs.
Standard organ dose calculations estimate tissue dose based on kerma in free-air at a reference location relative to the tissue.16 If available, Ka,r can be used as the basis of calculation; otherwise, phantom measurements are required to determine air-kerma levels. The calculation of skin dose from Ka,r requires consideration of several factors:
Consideration of these factors may be approached with varying degrees of sophistication ranging from a rough triage calculation to a complete skin dose map.18-20 For example, a triage may consist of an upper-limit calculation based on conservative assumptions that ignore dose spreading due to angulations (or separate beam angles on biplane systems). This simplified approach can save effort if the resulting conservative estimate is less than the sentinel event dose level.
Given that dose conversion factors and attenuation partially cancel one another, and that imaging geometries vary during procedures,21 a calibrated cumulative Ka,r dose indicator on a isocentric c-arm system (Fig. 2) might be expected to be nearly equal to or overestimate PSD. However, abnormal circumstances during a procedure or improper operating practices can lead to PSDs larger than the cumulative Ka,r. Therefore, the cumulative Ka,r must not be used as the sole indicator to rule out an FSE.
The uncertainty of a PSD calculation is difficult to quantify and can be large due to a number of compounding factors. Mapping fluoroscopy dose, in particular, is subject to large uncertainty because there are often no images to indicate where the dose was deposited. Varying the assumptions chosen regarding the fluoroscopy dose timeline can demonstrate this uncertainty range. Where possible, self-consistency checks should be applied to test the validity of fluoroscopy assumptions. For example, the time between runs will be well defined from the DICOM information and the fluoroscopic dose delivered in this time must be consistent with the dose rates expected for the equipment. Other factors which may not be well characterized—such as table movement, beam attenuation, or beam shaping filters—may also require multiple assumptions. Error contributions from machine-reported parameters can be relatively small (<10%), provided that the equipment is well characterized and that operational modes are known. Other small sources of uncertainty include the influence of beam quality, beam geometry, and body part shape on tissue and attenuation corrections.
If assumptions are consistently conservative, the net result will be to bias the PSD to higher values. In some cases, it may be possible to validate methods by the use of radiochromic film.22 In the absence of such validating data, the physicist must assign uncertainty to PSD estimates based on experience, particularly when near to the sentinel event threshold. Regardless, when fluoroscopic dose contributions predominate, the PSD estimate becomes more speculative in nature.
To demonstrate calculation methodologies, three cases involving varying degrees of information are presented.
Case 1: Bi-Plane C-arm With Run Images and Fluoroscopy Time Available
An older man with a partially thrombosed giant anterior communicating artery aneurysm presented for endovascular treatment. A diagnostic cerebral angiogram and interventional coiling procedure was performed on a bi-plane c-arm system not equipped with a dose monitor. The procedure was complicated by an arterial perforation, resulting in a cumulative fluoroscopy time of 154 minutes, which exceeded the investigational limit. Angiography comprised 22 frontal and lateral DSA runs (3 fps, ~20 frames each), and 1 rotational DSA run.
Staff interviews revealed that not all images were sent to PACS. Therefore, run images were obtained directly from the modality. DICOM information provided technique factors and patient-to-source distance for each run. Visual inspection of the image sets provided table translation and patient positioning information. Since the cumulative fluoroscopy time reported was the total for the frontal and lateral C-arm planes, the interviews also provided the fluoroscopy time split between the frontal and lateral arms, as well as details of the procedure timeline. An inspection of the modality verified the operating modes and display parameter calibrations.
Measurements using an anatomical phantom (Fig. 9) were performed to obtain output exposure and technique factors for fluoroscopy and angiographic acquisitions. Air kerma at the phantom surface was determined for the frontal and lateral planes with the phantom positioned at isocenter for a range of operating parameters. Measurements included the effects of attenuation through the table and padding.
Based on the information collected, a simplified approach was used to estimate PSD: C-arm angle variations were minor and imaging planes were close to perpendicular; therefore, maximum doses were calculated for the frontal and lateral fields instead of producing a dose map. Fluoroscopy time was split equally between the frontal and lateral planes and proportioned according to relative time gaps between runs. Fluoroscopy preceding each run was assumed to use the same positioning, geometry, magnification and collimation as the following DSA run. The DSA and rotational run air kerma measurements were scaled for patient position (1/r2) and technique factors (kVp2, mAs). Fluoroscopy exposures were corrected for SID and positioning. A conservative dose correction factor was applied.
The sum of the PSD for frontal and lateral fields did not exceed 15 Gy (Table 1); therefore, a detailed consideration of frontal-lateral field overlap was not required to rule out the occurrence of an FSE.
Fluoroscopy was performed at a default rate of 15 pulses per second (pps), so a recommendation was
made to lower the default rate to 7.5 pps. Consistent with the National Council on Radiation Protection (NCRP) Report 168 recommendation that potentially high dose procedures be performed only on fluoroscopy systems equipped with Ka,r monitors, a recommendation was made to upgrade the system with dose monitors.
Case 2: Single-Plane C-arm Equipped With an Air-Kerma Monitor
A middle-aged man with hyperlipidemia presented with exertional chest pain and a positive exercise tolerance test. A coronary angiogram and percutaneous revascularization procedure was performed using single plane fluoroscopy. The procedure was complicated by difficult catheterization and subsequent intracoronary thrombosis, resulting in complete occlusion of the left anterior descending (LAD) coronary artery and severe sustained coronary ischemia, which necessitated intubation with mechanical ventilation. The case cumulative Ka,r was 16 Gy from 168 minutes of fluoroscopy and 53 angiographic run acquisitions (15 fps, ~38 frames each), which triggered a PSD investigation.
Staff interviews provided information regarding fluoroscopy operation and timeline of the case. A modality inspection verified operating modes, air kerma meter calibration, and system geometry parameters. Public DICOM tag information for the runs included the kVp technique factors and image acquisition times. Because the contents of the tube current and dose-area-product tags were blank, x-ray output could not be calculated. Proprietary DICOM tags provided three dimensional table position.
Since only the run kVp technique factors were known, acrylic phantom measurements were performed to measure output air kerma for angiographic runs. Phantom thickness was varied to obtain dose dependence versus kVp for various magnification modes.
Assumptions and details of the calculation for this procedure included: A complete skin dose map was generated for a plane corresponding to the posterior surface of a supine patient lying on the table pad. Ka,r for each run was calculated based on phantom measurements. The total fluoroscopy dose was obtained by subtracting the sum of all run doses from the cumulative Ka,r. Fluoroscopy Ka,r was apportioned according to the relative time gaps between runs with consideration of the time spent intubating the patient. Fluoroscopy preceding each run was assumed to use the same positioning, geometry, magnification, and collimation as the following angiographic run. C-arm angulations and table position were included in the calculation on a frame-by-frame basis using DICOM information. Dose conversion and beam attenuation corrections were applied consistent with system geometry during each run.
Fig. 10 shows a map of the calculated skin dose at the patient’s back with a PSD of 9 Gy. In this case, the use of oblique views resulted in increased attenuation through the table and pad that numerically cancelled the increase in skin dose over Ka,r due to the backscatter and tissue conversion factors. Furthermore, using multiple angles spread the dose over non-overlapping entrance fields, and raising the table above the IRP decreased dose through 1/r2 reduction. The net result was a PSD estimate that was significantly lower than the Ka,r value. The uncertainty in this estimate was high because fluoroscopy comprised ~70% of the dose, but the use of consistently conservative assumptions throughout the calculation provided a reasonable degree of confidence that the true PSD was less than 15 Gy.
No operational recommendations were necessary, since best practice was used in a complicated case.
Case 3: Multiple Studies on a Bi-plane C-Arm Equipped with Air-Kerma Monitors
A middle-aged man with history of dural arteriovenous fistula underwent transvenous embolization of the fistula on a bi-plane c-arm system equipped with air-kerma monitors. The procedure was complicated by the tortuous and complex nature of the fistula. Fluoroscopy time and cumulative Ka,r for the case exceeded the investigational limits, and a dose estimate calculation was performed.
Cumulative Ka,r, PKA, and fluoroscopy time were available in the patient file on the modality. DICOM image sets provided run-by-run PKA that included preceding fluoroscopic dose and patient-to-source distance information. Staff interviews provided information regarding fluoroscopy operation and supported timeline details deduced from the DICOM information.
The accuracies of the dose monitor and the patient-to-source distance DICOM parameter were verified.
DICOM dose information was given in the form of PKA; however, DICOM collimator setting information was not available. Visual verification of run images was used to determine the area factor for air-kerma calculations. It was assumed that fluoroscopy was delivered with the same geometry as the angiographic run that followed it.
Similar to Case 1, the PSD for the case did not exceed 15 Gy for the summed frontal and lateral planes. A more detailed modeling was performed to determine if frontal and lateral field overlap was a factor; the results indicated significant overlap due to the c-arm angles used during the procedure.
Two months later, the patient returned for a second embolization procedure for the same fistula. Although the cumulative Ka,r and fluoroscopy time for the second procedure did not exceed the institutional threshold, an investigation was triggered on the basis of the sum of the two exam cumulative Ka,r values since the exams occurred within a 6 month window. The dose estimate indicated a cumulative PSD close to 15 Gy.
None, best practice was used in a complicated case. This patient did not exhibit signs of CRI. The two-month staging of the procedures may have decreased the potential for injury,5 but the case still fell within the 6-12 month Joint Commission time window. This example points out the need for a monitoring system to keep track of all procedures.
The fluoroscopic sentinel event is defined to be "prolonged fluoroscopy with a cumulative dose >1500 rads (15 Gy) to a single field". The definition of fluoroscopic sentinel event given by the Joint Commission employs the term "cumulative dose" in a different way than standard radiological physics usage. In the Joint Commission's meaning, this term refers to the PSD accumulated over a 6-12 month interval, not the cumulative reference point air-kerma (Ka,r) for a single-procedure that is required by the FDA on modern fluoroscopic equipment. An overall policy for detecting and investigating fluoroscopic sentinel events should be developed by each institution. In the event of a sentinel event, an institution must conduct a timely and thorough root cause analysis.
Substantial time and effort are required by the physicist to convert directly monitored parameters, such as fluoroscopy time or Ka,r, into estimated skin doses. Care should be given to the calculation, because overly conservative approaches will lead to an unnecessary and inappropriate root cause analysis. Even if the physicist has the best information regarding the procedure, significant errors can be associated with the dose estimate.
Arbique GM, Guild JB, Chason DP, Anderson JA. The Fluoroscopic Sentinel Event: What To Do?. J Am Osteopath Coll Radiol. 2014;3(3):8-20.