Journal of X-Ray Science and Technology 22 (2014) 309–319 DOI 10.3233/XST-140427 IOS Press
309
Impact of physician practice on patient radiation dose during CT guided biopsy procedures Victor J. Weira,∗ , Jie Zhangb and Angela P. Brunera a Baylor
Health Care System, Dallas, TX, USA of Radiology, University of Kentucky, Lexington, KY, USA
b Department
Received 2 May 2013 Revised 6 December 2013 Accepted 11 February 2014 Abstract. PURPOSE: Patient radiation dose during Computed Tomography (CT) guided biopsy procedures is determined by both acquisition technical parameters and physician practice. The potential effect of the physician practice is of concern. This study is to investigate the effects of those intangibles on patient radiation dose. METHODS: Patient radiation dose from 252 patients who underwent CT guided biopsy from 2009 to 2010 were retrospectively studied. Ten physicians who used conventional intermittent shots, low mA dose saving feature, or both were included in the study. The patient dose reports were retrieved and the total dose length products (DLPs) were analyzed. Linear regression analysis performed between various variables and reported dose. Patient detriment index (PDI) was developed, which sets threshold (standard of practice) for comparing physician practice with their peers. Odds ratio was calculated to determine odds of a group of patients receiving dose above threshold when compared to another group. RESULTS: Median DLP among ten physicians was 1194 mGy-cm. There was a significant difference (p < 0.01) between reported DLPs doses when physicians used dose saving feature vs. when feature not used (539.8 ± 169.4 mGy-cm vs. 1269.7 ± 659.0 mGy-cm). In general, physicians who used dose saving feature had lower relative PDIs (< 1) compared to the PDIs (> 1) without the dose feature. Odds ratio estimate of 7.7 at 95% confidence level indicates that the odds of a group receiving a high dose depends on practitioner. CONCLUSION: Adjustments of practice habits, use of dose saving features or both may be needed to improve patient care for CT biopsy. Keywords: CT fluoroscopy, radiation dose, CT dose, CT biopsy
1. Introduction The most commonly used imaging modalities for image-guided interventions and biopsies include x-ray computed tomography (CT), fluoroscopy, magnetic resonance imaging (MRI), and ultrasound (US). Fluoroscopy is used extensively in interventional procedures since it is real-time, lower cost and ∗
Corresponding author: Victor J. Weir, Texas Licensed Medical Physicist (Diagnostic Radiological Physics), Medical Physics & Radiation Safety, Baylor Health Care System, 4005 Crutcher Street, Suite 330, Dallas, TX 75246, USA. Tel.: +1 214 820 8516; E-mail: [email protected]. c 2014 – IOS Press and the authors. All rights reserved 0895-3996/14/$27.50
310
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
requires minimal post-processing or reconstruction. However, as a pure projection imaging technique, fluoroscopy may not be suitable for biopsy of some small lesions, especially when those lesions are close to major vessels [1]. Ultrasound also provides real-time imaging capabilities in a small form factor without the use of ionizing radiation. With a small probe, ultrasound can be easily handheld. Probes can also be inserted into body cavities or surgical fields for close access to the object being imaged. The problem with the use of a handheld US probe is that the probe must be pressed tightly against the skin and the arbitrary orientation of the probe results in different images from conventional crosssectional US images. Feasibility of magnetic resonance guided biopsy in patients was first reported by Stattaus et.al, using a wide bore short MRI system [2]. MRI images typically have excellent soft tissue contrast and good spatial resolution, so they are ideal for identifying structures and boundaries. However, image distortion is a potential complicating factor with MRI. Inhomogeneities in the magnetic field due to the patient and their carry-on devices can cause erroneous shifts in the image data. In addition the high magnetic field creates a very hazardous work environment. Therefore, MRI-based image-guided procedures pose unique challenges. CT provides excellent image quality for pretreatment imaging and treatment planning. With its ability to reconstruct images in three dimensions, CT can visualize the soft tissue especially the boundaries more clearly than with fluoroscopy where overlapping structures are compressed into a single plane, although the real-time performance of CT is inferior to fluoroscopy. With recent advances of CT techniques, CT fluoroscopy has become an indispensable tool for biopsy guidance. CT guidance for biopsies was first reported in 1976 by Haaga et al. [3], which used a conventional CT scan. The introduction of CT fluoroscopy by Katada et al. [4] allowed for faster image reconstruction, and nearly continuous image update, thus allowing for near real time image display. In situations where lesions are small and close to major vessels, CT fluoroscopy guidance is preferred [1]. The disadvantage of CT fluoroscopy is its potential for high patient radiation dose. To mitigate this issue, various features have been introduced to reduce CT radiation dose during interventional procedures, i.e., SmartStep (General Electric, Waukesha, Wis). Smartstep based CT fluoroscopy, is a feature that allows the use of low mA values and continuous real time scanning of a region as the needle is advanced to the target. When techniques such as SmartStep or other biopsy protocol are unavailable, conventional CT based on intermittent shots can be made to guide the physician or radiologist during biopsies. Intermittent shots, is the approach of taking single CT shots of the region of interest as the needle is advanced to the target. Typically, a routine CT protocol may be modified by reducing the mA or it may be used without modification. Thus, intermittent shots may involve a higher single exposure to the patient each time it is used. The number of intermittent shots taken influences how much dose the patient receives. Several articles published in the area of CT fluoroscopy biopsy guidance have focused on patient doses [5,6] or staff doses [7,8] or combinations of both [9]. One of the variables that can potentially impact patient doses is the operator/practitioner. However, to our knowledge, the practical variables, the operator and practitioner that can potentially impact patient doses have not been studied until now. In this extensive study, we attempt to answer the question, how does physician practice influence patient doses? One practical variable involves how much noise physicians are willing to accept in each image they view, a choice that influences how much mA was used for scanning the patient, especially when conventional CT (intermittent shots) were used. Another variable of physician practice involves whether to use a CT biopsy feature such as SmartStep or not. If a physician chooses not to use SmartStep, then they have, by default, chosen to use conventional CT. This study is not about comparing SmartStep to conventional CT. It is about the impact that a physician’s choice of SmartStep or conventional CT will have on the scanner displayed doses for patients under their care. We demonstrate this impact by introducing the concept of a Patient Detriment Index (PDI), and also the odds ratio calculation.
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
311
2. Methods We retrospectively retrieved 381 patients who underwent CT guided interventional and biopsy procedures at Baylor University Medical Center from November 2009 to November 2010. 252 patient data were analyzed, including 26 patient procedures where SmartStep was used by those physicians who had also performed procedures without using Smartstep (n = 23). 129 patients were excluded due to their incomplete date sets, particularly unavailable dose report. The study was approved by the IRB at the Baylor Healthcare System. 2.1. Data acquisition Ten physicians were involved in all interventional procedures during the period studied. To protect their identities they were labeled with the letters of the alphabet (A – J). At our institution, a GE lightspeed PRO 16 was used for all the biopsies performed. Typically, technique parameters used for intermittent CT shots during biopsies and interventional guidance were initially set at 120 kVp, 200 mA, and 1.375 of pitch. For intermittent shots, the mA can be changed depending on the physician preference. The number of intermittent shots required to complete the procedure was also a variable since individual physician required different numbers of CT shots. Intermittent shots were for tube rotation times of 1 second. If SmartStep was used, the techniques were 120 kVp and 40 mA. SmartStep is an optional package on the GE CT scanner for interventional procedures. With the package comes an in room monitor, a handheld remote control, and a foot switch for making exposures. This package allows for in room review of images and scan initiation and control. After identifying CT biopsy procedures, the patient dose report was analyzed in the Picture Archiving and Communication System (PACS) (GE CentricityTM ) for the number of CT intermittent shots taken, as well as the total Dose Length Product (DLP) (mGycm) displayed for the study. The physician and type of procedure were identified by looking through the exam notes. 2.2. Data analysis A preliminary analysis was done to test for normality by plotting the histograms and fitting to a normal distribution. Following those tests, we focused on the data from two radiologists in the group, E and H. For data analysis, first we compared two physicians who used only conventional CT based on interc 2007 mittent shots. Data was analyzed using the descriptive statistics feature in Microsoft ExcelTM ( Microsoft Corporation, Washington). Second we compared physicians who used SmartStep vs. those who did not use SmartStep. Third we excluded physicians who never used SmartStep, and analyzed the results of those who used SmartStep for some of their procedures, and compared their DLPs for when they used SmartStep vs when they used conventional CT shots. 2.3. Conventional CT intermittent shots First a linear regression analysis was performed to investigate any possible relationship between the variable (physician performing the biopsy) and the DLP displayed for patients. Second, for data with normal distribution, several tests concerning number of CT intermittent shots and total DLP were performed. The two physicians that performed the most procedures in our institution over the one year period were analyzed. Data from one modeled a normal distribution (E, n = 56) and the other (H, n = 147) was a model for a highly skewed distribution that was not normal. For a period of six months, we
312
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures Table 1 Common biopsy procedures, average DLP and intermittent CT shots for physicians H and E
Pulmonary mass biopsy Lung core biopsy Lung nodule biopsy CT biopsy Average
Average DLP (mGy-cm) 1408.30 1250.10 1648.00 1926.30 1558.18
H Intermittent CT shots (sec) 8.25 11.86 11.00 17.17 12.07
Average DLP (mGy-cm) 651.14 758.60 810.90 661.00 720.41
E Intermittent CT shots (sec) 8.57 10.70 13.00 7.40 9.92
Table 2 Different biopsy procedures, average DLP and intermittent CT shots for physician H and E H E Average Intermittent # of Average Intermittent # of DLP CT shots cases DLP CT shots cases (mGy-m) (sec) (mGy-cm) (sec) Drainage 979.30 6.60 11 Psoas mass biopsy 749.40 10.00 3 Drain Placement, Abdomen 1327.00 9.00 5 Lymph node biopsy 965.20 11.50 2 Abdominal psoas aspiration/drainage 1639.30 9.00 3 Retroperitoneal Biopsy 603.20 6.33 3 Hepatic lesion core biopsy 1539.17 20.00 1 Lung Aspiration 253.00 3.00 1 Lung biopsy, fiducial placement 4320.45 22.00 1 Bone/iliac biopsy 793.00 9.30 6 Average 1961.04 13.32 4.2 Average 672.8 8.02 3
analyzed the biopsies that both physicians had in common. For physician E, 24% (11/45) of cases were lung biopsies. For physician H, 20.9% (14/67) of cases were lung biopsies. Further breakdown of the lung biopsy data is shown in Table 1. Besides these, there were other procedures performed that are also shown in Table 2. Data was also collected on average mA and average patient diameter for a period of four months for both physicians. The patient dimensions were taken to investigate if there was any justification for using different mA values for the patient groups of physicians E and H. Anterior (AP) and lateral (LAT) patient dimensions were collected from patient scout measurements. A geometric mean was calculated as the square root of (AP*LAT). 2.4. SmartStep use vs. Conventional CT Next we computed the overall data for all physicians whether they used SmartStep (n = 26) or not (n = 226). We hypothesized that there is no difference between doses with /without the use of the SmartStep feature. 2.5. Influence of SmartStep experience To investigate possible influence of SmartStep experience, we separated the data for those physicians that used SmartStep for some of their biopsies into two groups. One group was for procedures done with SmartStep (n = 26), the other was procedures those physicians did without SmartStep (n = 23). We then average the results per group for total DLP (mGy-cm) and times (s) for CT fluoroscopy or intermittent shots. The overall data was analyzed using parametric methods. We computed the median DLP value for the entire group of physicians, as well as medians for number of intermittent shots (seconds). Data on the years of experience each physician has in regards to performing CT biopsies was also obtained.
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
313
2.6. Patient Detriment Index (PDI) Patient Detriment Index (PDI) was developed by taking the median data for the entire group of physicians and using it to normalize the average DLP data for each physician. We did not weigh higher doses against the potential benefit to the patient, as a result of perhaps good image quality or successful completion of the biopsy procedure. The median DLP was therefore set as the threshold for developing the PDI. With this approach, we can use the median DLP as the standard of practice, and any DLPs above the median DLP will be considered a detriment to patients while DLPs below will be desirable. 2.7. Odds ratio calculation Finally, to put all these data into perspective, we include a calculation of the odds ratio; which is the ratio of the odds of an event occurring in one group to the odds of it occurring in another group. The odds ratio is used to directly compare the patient groups of the two physicians, E and H, in terms of the odds of a patient in one group receiving a DLP above the median DLP threshold when treated by either physician H (“treatment” group) or Physician E (“control” group). An odds ratio of 1 indicates that the odds of a patient receiving a DLP above the threshold value is equally likely for patients in both groups. An odds ratio greater than 1 indicates that a DLP above the threshold is more likely to occur in the “treatment” group than in the “control” group, and an odds ratio less than 1 indicates that a DLP above the threshold is less likely to occur in the “treatment” group than in the “control” group. Our assumption is that the patients treated by physician E can be considered the “control” in an odds ratio calculation since the average DLP for patients treated by physician E were below the threshold for the entire group. Patients treated by physician H were considered the “treatment” group. 3. Results 3.1. Influence of physician experience We collected the experience of each physician and plotted it against their average DLP values to determine any correlation between experience and physician performance. The result is reported in Fig. 1. The correlation between mean DLPs (mGy-cm) and experience (years) was 0.0226. 3.2. Conventional CT intermittent shots Our statistic results show that there is a significant difference between the doses from physician E and H. The average total DLPs calculated for both physicians E and H were 720.4 ± 77.4 mGy-cm, and 1558.2 ± 294.9 mGy-cm respectively for the same type of biopsies in Table 1. The average patient diameter for both physicians was obtained from calculation of the geometric mean of the AP and LAT dimensions. For physician E, patient average diameter was 294.2 mm, and for physician H, average patient diameter was 296.3 mm. The average mA for both physicians were (210 ± 35 mA), and (289 ± 50.9 mA) for E and H, respectively. These physicians E and H required on the average 9.9 ± 2.5 s, and 12.1 ± 3.7 s intermittent shots respectively for the same type of biopsies in Table 1. Each shot is at 1 second. Table 2 also shows different procedures performed by E and H. H required on average 1961.04 mGy-cm DLP compared to 672.8 mGy-cm DLP on average for E, for cases performed. Linear regression analysis result, Fig. 2, for 10 physicians. A relationship between the number of intermittent shots needed to complete a procedure and the patient DLP is demonstrated.
314
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
Fig. 1. Physician experience (years) vs. average Dose-Length Product (DLP) mGy-cm. Numbers next to each data point represent the years of experience for that physician.
Fig. 2. Linear regression analysis showing the number of intermittent CT shots required to complete a procedure vs. the patient DLP (mGy-cm). Physician labels and years of experience are also shown.
3.3. SmartStep use vs. Conventional CT When the use of Smartstep vs non-use of SmartStep was analyzed for the entire physician group, our statistic results show that there is a significant difference (p < 0.01) between the doses patients received (DLP = 539.8 ± 169.4 mGy-cm) when physicians used SmartStep vs. when Smartstep was not used (DLP = 1269.7 ± 659 mGy-cm).
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
315
Fig. 3. Average DLP per physician practice, showing the total number of procedures each physician performed. These numbers are for both procedures with and without Smartstep where relevant.
3.4. Influence of SmartStep experience When we take out the data for physicians who have never used SmartStep, and focus only on those physicians who have used SmartStep at some time in their practice, we find an interesting result. When Smartstep was used (n = 26), the average patient dose (DLP) was 539.8 ± 169.4 mGy-cm. When Smartstep was not used (n = 23), the average patient dose was 1133 ± 349 mGy-cm. This implies that for those physicians who used smartstep for some procedures and did not use it for others, there was a significant difference (p < 0.01) in their practice in regards to patient dose. For three physicians (I, G, and D) who used smartstep for some procedures and did not use it for others, the doses patients received were very comparable (Fig. 3). While physicians C & J performed procedures where the patient doses were on the high end leading to the high average for this group. 3.5. Patient Detriment Index (PDI) The average DLP (mGy-cm) was compared for each physician. The ratio of the total average DLP associated with each physician was normalized against the median DLP value for the group. As shown in Fig. 4, of the six physicians with a PDI < 1, four have used SmartStep. Of the four physicians with PDI > 1, one (C) has used the SmartStep feature. Of the 9 biopsies performed by physician C, 3 involved smartstep at significantly lower doses (DLP = 346 mGy-cm) than the 6 without smartstep (DLP = 2426
316
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
Fig. 4. Patient Detriment Index (PDI) values are shown above with the number of procedures each physician performed. The solid black line represents the PDI threshold value of 1. PDI less than one indicates good practice while PDI more than one is considered to be detrimental to the patient.
mGy-cm). Furthermore, all those physicians who used SmartStep, except C, have a lower relative PDI compared to their PDI without SmartStep. Three of five physicians who did not use SmartStep, have PDIs > 1 (Fig. 4). 3.6. Odds ratio calculation The odds ratio is 7.7 with a 95% confidence level and a range of 3.26 to 18.10. 4. Discussion Physician practice during CT guidance of biopsy plays a major role in determining patient radiation dose. When physicians E and H were compared for the use of conventional CT intermittent shots, a significant difference was found between the dose to patients and the number of intermittent shots taken. Both physicians did not use SmartStep during procedures for the time period analyzed. However, physician E had consistently lower doses than physician H. Doses were 720.4 ± 77.4 mGy-cm vs. 1558.2 ± 294.9 mGy-cm on the average for physicians E and H respectively for lung biopsy data analyzed over a six month period. These numbers can be partly explained by the fact that physician H required higher mA values (289 ± 50.9 mA) compared to physician E (210 ± 35 mA). In addition, physician E required fewer CT intermittent shots; 9.9 ± 2.5 s compared to physician H, who required 12.1 ± 3.7 s intermittent shots to complete lung biopsy studies during the six month period considered. The choice of a combination of more intermittent shots and higher mA requirements will lead to higher patient doses
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
317
for H compared to E. One plausible explanation of the difference between these physicians is their expectations of image quality. This is evidenced by the fact that the average patient dimensions for the both physicians were identical. For physician E, patient average diameter was 294.2 mm, and for physician H, average patient diameter was 296.3 mm. We would therefore expect that image noise for the two patient populations would not be far apart, so the use of more mA (289 ± 50.9 mA) by H compared to E (210 ± 35 mA) may be purely a preference for image appearance. We also expected that a procedure’s degree of difficulty may play a role in how many intermittent shots was required for completion. Upon analysis of various random procedures performed by both physicians, we found the performance of the two physicians to be consistent with the data in Table 1. The general trend for these procedures was the same for both E and H, with H requiring in general higher mA and longer times to complete procedures. This result is seen in Table 2 where a wide variety of procedures are shown, but with one general trend, and that is the difference in DLP and number of shots needed to complete the procedure. Differences between procedures with and without the use of SmartStep data were significant. In analysing the data for SmartStep use, our results revealed that when SmartStep was used by those physicians who used Smartstep some of the time, those physicians had high doses (DLP = 1133 mGy-cm) when they did not use SmartStep. These high doses were comparable to the doses patients received (DLP = 1269.75 mGy-cm) when they were treated without the use of SmartStep. One would expect that physicians who have used a dose savings feature would be more sensitive to reducing dose when they used intermittent shots than their peers who had never used a dose savings feature. This sensitivity to reducing patient dose would be reflected in the willingness to use a lower mA for intermittent shots when they did not use the dose savings feature. The fact that the data for the two groups is comparable would suggest that physicians need to be more proactive when it comes to scanning parameters used during their biopsy procedures. However, because of the small sample size, this average data could be misleading. For three Physicians I, G, D, the results were quite interesting. Physicians I, G, D used Smartstep for some procedures and did not use smartstep for other procedures. Their results show that the patient doses were not significantly different whether they used Smartstep or not. This is a desirable result showing that these physicians are mindful about using low dose techniques whether they used Smartstep or not. While this is encouraging, more data is needed to reach a solid conclusion about the practice for this group of physicians who used Smartstep for some procedures and not for others. It must be pointed out that physicians C & J performed procedures where the patient doses were on the high end leading to the high average for this group. We also considered the possibility that physician experience may play a role in the dose patients received when biopsies were performed. It is expected that a less experienced physician may use more radiation in order to have a better image quality for performing the biopsy. Figure 1 shows that there was no strong correlation between experience and the DLP values (R2 = 0.0226). Currently vendors provide dose savings features on their scanner but not all physician take advantage of these features. This leads to large variations in dose to patients. Our results show that this feature can play a large role in reducing dose to patients, and can potentially minimize large variations in patient dose caused by physician practice or preference. When comparing use vs non-use of SmartStep, it must be pointed out that the data was not collected on the same patient during a given procedure, but on different patients. While for different patients, the same procedure may be done over longer times and experience different complications, with a large enough dataset of patients, we expect such variations to be smoothed out. Nevertheless, we can still see a difference between doses and times for the use of dose savings or non-use of dose savings features. We compared the average total DLP (mGy-cm) for each physician with the median total DLP for the entire group over the one year period. The ratio of the total DLP associated with each physician was
318
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
normalized against the median value and called the patient detriment index (PDI). This index compares by how much each physician deviates from the entire group, and can serve as a way to rank physicians by their practice. Since the data was only approximately normally distributed, it was more appropriate to use the median values instead of the mean values. The mean in this situation will not represent the central tendency of the distribution. For data that has a Gaussian distribution, the average will be equal to the median, so using the average or the median will lead to the same result. In Fig. 3, It can be seen that although some physicians (E, A) did not use smartstep, their PDIs were still lower than 1. On the other hand, C used smartstep but had a PDI>1. C performed 9 studies during the time period analyzed. Of this number 3 were with Smartstep where the doses were low, and 6 without Smartstep. The 6 without Smartstep had the highest doses among the group of physicians. The utility of PDI is that it reveals the difference between physicians in the group. Since SmartStep saves dose, not using it may lead to a higher PDI compared to those who use SmartStep. Even if all physicians used SmartStep, PDIs will be a way to compare physicians with their peers in the group since there will be a lower median dose threshold to use as a benchmark, and all physicians will be compared to this lower threshold number. This will help drive doses even lower. In short, the use of a PDI could help rank physicians according to their practice. Healthcare organization could use these rankings to make improvements by defining their own patient detriment index. Also, the PDI can be applied to any health care center to allow physicians to minimize radiation dose by providing a metric or guidelines related to their practice in that institution relative to the standard of practice in the community. Ultimately, it could be used to compare practices at various institutions and could help to develop a national standard once more data from various institutions is available. With full implementation of the Affordable Care Act in 2014, the PDI could be included in patient care metrics to demonstrate how healthcare providers are working to provide the best possible care to patients. It can be seen how a concept like the PDI or similar ideas could be implemented in medical Imaging departments nationwide. Another potential benefit of using the PDI is to set a threshold dose for the purpose of Joint commision sentinel dose events monitoring in CT guided interventional procedures. Currently, with all the attention that fluroscopy is receiving, the time has come to set thresholds not only for CT guided interventions but also for CT procedures in general. The odds ratio was used to indicate the likelihood of patients receiving a DLP above the median DLP threshold value or below the threshold value. The odds ratio for a DLP above the threshold value was 7.7. Since this number was larger than 1, we concluded that patients in the “treatment” group (group treated by physician H), were more likely than patients in the “control” group (treated by physician E) to receive a DLP above the threshold value. The odds ratio is calculated with a 95% confidence interval, and a range from 3.23 to 18.10. In other words, patients treated by H have their risks of receiving a dose above the threshold value increased by 7.7 times, with a range of 3.26 to 18.1, as compared to those treated by E. Lastly, this study is very important to provide information to staff, and to provide guidelines on training staff to minimize radiation dose to patients during CT guided interventional procedures. Potential guidelines include, using dose savings features available for biopsies, using lower mA values for intermittent CT shots, and also using the fewest number of CT shots possible to complete a procedure. A limitation of this study could be the wide variation between physicians on the number and types of procedures performed. However, we believe this limitation is not a detriment to our study, it being a study that can provide practice guideline for physicians and staff on ways to reduce patient doses. Another limitation is the limited data set of procedures performed by some physicians. A larger multi-hospital
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
319
study involving a larger number of physicians over a few years may be necessary to set a standard of practice for a specific procedure. For the odds ratio calculation, the assignment of a “treatment” group, as patients treated by physician H, or a “control” group, as patients treated by physician E can be questioned. We believe our application provides a valid approximation since we have set a threshold value for DLP that we expect can be used a a benchmark for “treatment”: when patients receive a dose above this threshold, vs “control”, when patients receive doses below this threshold value. To our knowledge, this study presents for the first time the concept of the PDI as a way of comparing the practice of physicians with their peers. We also present for the first time the use of the odds ratio in the context of medical practice. 5. Conclusion Scanner reported patient doses can be significantly influenced by physician practice during CT guided biopsy procedures. The potential impact may be mitigated by the use of the following strategies. First, a dedicated biopsy protocol should be created, based on right practice, right dose and best outcome as outlined by the American College of Radiology. Second, physicians should use dose savings features such as SmartStep when available. Third, if intermittent shots are used, lower mA values and fewer intermittent shots should be used. With the current push to record patient doses on interventional procedures, setting up a local dose threshold based on physician practice in an institution can help healthcare providers identify areas where improvements need to be made with regards to lowering patient dose. Such improvements can be made by combinations of practice guidelines, staff training, equipment upgrades, and incentives, among others. References [1] [2] [3] [4] [5] [6] [7] [8] [9]
S. Sheth, U.M. Hamper, D.B. Stanley et al., US Guidance for Thoracic Biopsy: A Valuable Alternative to CT, Radiology 210 (1999), 726–726. J. Stattaus, S. Maderwald, M. Forsting et al., MR guided core biopsy with MR fluoroscopy using a short, wide-bore 1.5-Tesla scanner: feasibility and initial results, J Magn Reson Imaging 27 (2008), 1181–1187. J.R. Haaga and R.J. Alfidi, Precise biopsy localization by computer tomography, Radiology 118 (1976), 603–607. K. Katada, H. Anno, Y. Ogura et al., Development and early trials of real-time CT fluoroscopy. Neuroradiology 37 (1995), 587–588. S. Silverman, K. Tuncali, D. Adams et al., CT Fluoroscopy-guided Abdominal Interventions: Techniques, Results, and Radiation Exposure, Radiology 212 (1999), 673–681. V. Tsapaki, C. Triantopoulou, P. Maniatis et al., Patient Skin Dose assessment during CT-guided interventional procedures, Radiation Protection Dosimetry 129(1–3) (2008), 29–31. E. Paulson, D. Sheafor, D. Enterline et al., CT Fluoroscopy – guided Interventional Procedures: Techniques and Radiation Dose to Radiologists, Radiology 220 (2001), 1612–1617. B. Daly, T.L. Krebs, J.J. Wong-You-Cheong et al., Percutaneous Abdominal and Pelvic Interventional Procedures Using CT Fluoroscopy Guidance, AJR 173 (1999), 637–644. W. Teeuwisse, J. Geleijns, J. Broerse et al., Patient and staff dose during CT guided biopsy, drainage and coagulation, The British Journal of Radiology 74 (2001), 720–726.
Copyright of Journal of X-Ray Science & Technology is the property of IOS Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
Copyright of Journal of X-Ray Science & Technology is the property of IOS Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
309
Impact of physician practice on patient radiation dose during CT guided biopsy procedures Victor J. Weira,∗ , Jie Zhangb and Angela P. Brunera a Baylor
Health Care System, Dallas, TX, USA of Radiology, University of Kentucky, Lexington, KY, USA
b Department
Received 2 May 2013 Revised 6 December 2013 Accepted 11 February 2014 Abstract. PURPOSE: Patient radiation dose during Computed Tomography (CT) guided biopsy procedures is determined by both acquisition technical parameters and physician practice. The potential effect of the physician practice is of concern. This study is to investigate the effects of those intangibles on patient radiation dose. METHODS: Patient radiation dose from 252 patients who underwent CT guided biopsy from 2009 to 2010 were retrospectively studied. Ten physicians who used conventional intermittent shots, low mA dose saving feature, or both were included in the study. The patient dose reports were retrieved and the total dose length products (DLPs) were analyzed. Linear regression analysis performed between various variables and reported dose. Patient detriment index (PDI) was developed, which sets threshold (standard of practice) for comparing physician practice with their peers. Odds ratio was calculated to determine odds of a group of patients receiving dose above threshold when compared to another group. RESULTS: Median DLP among ten physicians was 1194 mGy-cm. There was a significant difference (p < 0.01) between reported DLPs doses when physicians used dose saving feature vs. when feature not used (539.8 ± 169.4 mGy-cm vs. 1269.7 ± 659.0 mGy-cm). In general, physicians who used dose saving feature had lower relative PDIs (< 1) compared to the PDIs (> 1) without the dose feature. Odds ratio estimate of 7.7 at 95% confidence level indicates that the odds of a group receiving a high dose depends on practitioner. CONCLUSION: Adjustments of practice habits, use of dose saving features or both may be needed to improve patient care for CT biopsy. Keywords: CT fluoroscopy, radiation dose, CT dose, CT biopsy
1. Introduction The most commonly used imaging modalities for image-guided interventions and biopsies include x-ray computed tomography (CT), fluoroscopy, magnetic resonance imaging (MRI), and ultrasound (US). Fluoroscopy is used extensively in interventional procedures since it is real-time, lower cost and ∗
Corresponding author: Victor J. Weir, Texas Licensed Medical Physicist (Diagnostic Radiological Physics), Medical Physics & Radiation Safety, Baylor Health Care System, 4005 Crutcher Street, Suite 330, Dallas, TX 75246, USA. Tel.: +1 214 820 8516; E-mail: [email protected]. c 2014 – IOS Press and the authors. All rights reserved 0895-3996/14/$27.50
310
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
requires minimal post-processing or reconstruction. However, as a pure projection imaging technique, fluoroscopy may not be suitable for biopsy of some small lesions, especially when those lesions are close to major vessels [1]. Ultrasound also provides real-time imaging capabilities in a small form factor without the use of ionizing radiation. With a small probe, ultrasound can be easily handheld. Probes can also be inserted into body cavities or surgical fields for close access to the object being imaged. The problem with the use of a handheld US probe is that the probe must be pressed tightly against the skin and the arbitrary orientation of the probe results in different images from conventional crosssectional US images. Feasibility of magnetic resonance guided biopsy in patients was first reported by Stattaus et.al, using a wide bore short MRI system [2]. MRI images typically have excellent soft tissue contrast and good spatial resolution, so they are ideal for identifying structures and boundaries. However, image distortion is a potential complicating factor with MRI. Inhomogeneities in the magnetic field due to the patient and their carry-on devices can cause erroneous shifts in the image data. In addition the high magnetic field creates a very hazardous work environment. Therefore, MRI-based image-guided procedures pose unique challenges. CT provides excellent image quality for pretreatment imaging and treatment planning. With its ability to reconstruct images in three dimensions, CT can visualize the soft tissue especially the boundaries more clearly than with fluoroscopy where overlapping structures are compressed into a single plane, although the real-time performance of CT is inferior to fluoroscopy. With recent advances of CT techniques, CT fluoroscopy has become an indispensable tool for biopsy guidance. CT guidance for biopsies was first reported in 1976 by Haaga et al. [3], which used a conventional CT scan. The introduction of CT fluoroscopy by Katada et al. [4] allowed for faster image reconstruction, and nearly continuous image update, thus allowing for near real time image display. In situations where lesions are small and close to major vessels, CT fluoroscopy guidance is preferred [1]. The disadvantage of CT fluoroscopy is its potential for high patient radiation dose. To mitigate this issue, various features have been introduced to reduce CT radiation dose during interventional procedures, i.e., SmartStep (General Electric, Waukesha, Wis). Smartstep based CT fluoroscopy, is a feature that allows the use of low mA values and continuous real time scanning of a region as the needle is advanced to the target. When techniques such as SmartStep or other biopsy protocol are unavailable, conventional CT based on intermittent shots can be made to guide the physician or radiologist during biopsies. Intermittent shots, is the approach of taking single CT shots of the region of interest as the needle is advanced to the target. Typically, a routine CT protocol may be modified by reducing the mA or it may be used without modification. Thus, intermittent shots may involve a higher single exposure to the patient each time it is used. The number of intermittent shots taken influences how much dose the patient receives. Several articles published in the area of CT fluoroscopy biopsy guidance have focused on patient doses [5,6] or staff doses [7,8] or combinations of both [9]. One of the variables that can potentially impact patient doses is the operator/practitioner. However, to our knowledge, the practical variables, the operator and practitioner that can potentially impact patient doses have not been studied until now. In this extensive study, we attempt to answer the question, how does physician practice influence patient doses? One practical variable involves how much noise physicians are willing to accept in each image they view, a choice that influences how much mA was used for scanning the patient, especially when conventional CT (intermittent shots) were used. Another variable of physician practice involves whether to use a CT biopsy feature such as SmartStep or not. If a physician chooses not to use SmartStep, then they have, by default, chosen to use conventional CT. This study is not about comparing SmartStep to conventional CT. It is about the impact that a physician’s choice of SmartStep or conventional CT will have on the scanner displayed doses for patients under their care. We demonstrate this impact by introducing the concept of a Patient Detriment Index (PDI), and also the odds ratio calculation.
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
311
2. Methods We retrospectively retrieved 381 patients who underwent CT guided interventional and biopsy procedures at Baylor University Medical Center from November 2009 to November 2010. 252 patient data were analyzed, including 26 patient procedures where SmartStep was used by those physicians who had also performed procedures without using Smartstep (n = 23). 129 patients were excluded due to their incomplete date sets, particularly unavailable dose report. The study was approved by the IRB at the Baylor Healthcare System. 2.1. Data acquisition Ten physicians were involved in all interventional procedures during the period studied. To protect their identities they were labeled with the letters of the alphabet (A – J). At our institution, a GE lightspeed PRO 16 was used for all the biopsies performed. Typically, technique parameters used for intermittent CT shots during biopsies and interventional guidance were initially set at 120 kVp, 200 mA, and 1.375 of pitch. For intermittent shots, the mA can be changed depending on the physician preference. The number of intermittent shots required to complete the procedure was also a variable since individual physician required different numbers of CT shots. Intermittent shots were for tube rotation times of 1 second. If SmartStep was used, the techniques were 120 kVp and 40 mA. SmartStep is an optional package on the GE CT scanner for interventional procedures. With the package comes an in room monitor, a handheld remote control, and a foot switch for making exposures. This package allows for in room review of images and scan initiation and control. After identifying CT biopsy procedures, the patient dose report was analyzed in the Picture Archiving and Communication System (PACS) (GE CentricityTM ) for the number of CT intermittent shots taken, as well as the total Dose Length Product (DLP) (mGycm) displayed for the study. The physician and type of procedure were identified by looking through the exam notes. 2.2. Data analysis A preliminary analysis was done to test for normality by plotting the histograms and fitting to a normal distribution. Following those tests, we focused on the data from two radiologists in the group, E and H. For data analysis, first we compared two physicians who used only conventional CT based on interc 2007 mittent shots. Data was analyzed using the descriptive statistics feature in Microsoft ExcelTM ( Microsoft Corporation, Washington). Second we compared physicians who used SmartStep vs. those who did not use SmartStep. Third we excluded physicians who never used SmartStep, and analyzed the results of those who used SmartStep for some of their procedures, and compared their DLPs for when they used SmartStep vs when they used conventional CT shots. 2.3. Conventional CT intermittent shots First a linear regression analysis was performed to investigate any possible relationship between the variable (physician performing the biopsy) and the DLP displayed for patients. Second, for data with normal distribution, several tests concerning number of CT intermittent shots and total DLP were performed. The two physicians that performed the most procedures in our institution over the one year period were analyzed. Data from one modeled a normal distribution (E, n = 56) and the other (H, n = 147) was a model for a highly skewed distribution that was not normal. For a period of six months, we
312
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures Table 1 Common biopsy procedures, average DLP and intermittent CT shots for physicians H and E
Pulmonary mass biopsy Lung core biopsy Lung nodule biopsy CT biopsy Average
Average DLP (mGy-cm) 1408.30 1250.10 1648.00 1926.30 1558.18
H Intermittent CT shots (sec) 8.25 11.86 11.00 17.17 12.07
Average DLP (mGy-cm) 651.14 758.60 810.90 661.00 720.41
E Intermittent CT shots (sec) 8.57 10.70 13.00 7.40 9.92
Table 2 Different biopsy procedures, average DLP and intermittent CT shots for physician H and E H E Average Intermittent # of Average Intermittent # of DLP CT shots cases DLP CT shots cases (mGy-m) (sec) (mGy-cm) (sec) Drainage 979.30 6.60 11 Psoas mass biopsy 749.40 10.00 3 Drain Placement, Abdomen 1327.00 9.00 5 Lymph node biopsy 965.20 11.50 2 Abdominal psoas aspiration/drainage 1639.30 9.00 3 Retroperitoneal Biopsy 603.20 6.33 3 Hepatic lesion core biopsy 1539.17 20.00 1 Lung Aspiration 253.00 3.00 1 Lung biopsy, fiducial placement 4320.45 22.00 1 Bone/iliac biopsy 793.00 9.30 6 Average 1961.04 13.32 4.2 Average 672.8 8.02 3
analyzed the biopsies that both physicians had in common. For physician E, 24% (11/45) of cases were lung biopsies. For physician H, 20.9% (14/67) of cases were lung biopsies. Further breakdown of the lung biopsy data is shown in Table 1. Besides these, there were other procedures performed that are also shown in Table 2. Data was also collected on average mA and average patient diameter for a period of four months for both physicians. The patient dimensions were taken to investigate if there was any justification for using different mA values for the patient groups of physicians E and H. Anterior (AP) and lateral (LAT) patient dimensions were collected from patient scout measurements. A geometric mean was calculated as the square root of (AP*LAT). 2.4. SmartStep use vs. Conventional CT Next we computed the overall data for all physicians whether they used SmartStep (n = 26) or not (n = 226). We hypothesized that there is no difference between doses with /without the use of the SmartStep feature. 2.5. Influence of SmartStep experience To investigate possible influence of SmartStep experience, we separated the data for those physicians that used SmartStep for some of their biopsies into two groups. One group was for procedures done with SmartStep (n = 26), the other was procedures those physicians did without SmartStep (n = 23). We then average the results per group for total DLP (mGy-cm) and times (s) for CT fluoroscopy or intermittent shots. The overall data was analyzed using parametric methods. We computed the median DLP value for the entire group of physicians, as well as medians for number of intermittent shots (seconds). Data on the years of experience each physician has in regards to performing CT biopsies was also obtained.
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
313
2.6. Patient Detriment Index (PDI) Patient Detriment Index (PDI) was developed by taking the median data for the entire group of physicians and using it to normalize the average DLP data for each physician. We did not weigh higher doses against the potential benefit to the patient, as a result of perhaps good image quality or successful completion of the biopsy procedure. The median DLP was therefore set as the threshold for developing the PDI. With this approach, we can use the median DLP as the standard of practice, and any DLPs above the median DLP will be considered a detriment to patients while DLPs below will be desirable. 2.7. Odds ratio calculation Finally, to put all these data into perspective, we include a calculation of the odds ratio; which is the ratio of the odds of an event occurring in one group to the odds of it occurring in another group. The odds ratio is used to directly compare the patient groups of the two physicians, E and H, in terms of the odds of a patient in one group receiving a DLP above the median DLP threshold when treated by either physician H (“treatment” group) or Physician E (“control” group). An odds ratio of 1 indicates that the odds of a patient receiving a DLP above the threshold value is equally likely for patients in both groups. An odds ratio greater than 1 indicates that a DLP above the threshold is more likely to occur in the “treatment” group than in the “control” group, and an odds ratio less than 1 indicates that a DLP above the threshold is less likely to occur in the “treatment” group than in the “control” group. Our assumption is that the patients treated by physician E can be considered the “control” in an odds ratio calculation since the average DLP for patients treated by physician E were below the threshold for the entire group. Patients treated by physician H were considered the “treatment” group. 3. Results 3.1. Influence of physician experience We collected the experience of each physician and plotted it against their average DLP values to determine any correlation between experience and physician performance. The result is reported in Fig. 1. The correlation between mean DLPs (mGy-cm) and experience (years) was 0.0226. 3.2. Conventional CT intermittent shots Our statistic results show that there is a significant difference between the doses from physician E and H. The average total DLPs calculated for both physicians E and H were 720.4 ± 77.4 mGy-cm, and 1558.2 ± 294.9 mGy-cm respectively for the same type of biopsies in Table 1. The average patient diameter for both physicians was obtained from calculation of the geometric mean of the AP and LAT dimensions. For physician E, patient average diameter was 294.2 mm, and for physician H, average patient diameter was 296.3 mm. The average mA for both physicians were (210 ± 35 mA), and (289 ± 50.9 mA) for E and H, respectively. These physicians E and H required on the average 9.9 ± 2.5 s, and 12.1 ± 3.7 s intermittent shots respectively for the same type of biopsies in Table 1. Each shot is at 1 second. Table 2 also shows different procedures performed by E and H. H required on average 1961.04 mGy-cm DLP compared to 672.8 mGy-cm DLP on average for E, for cases performed. Linear regression analysis result, Fig. 2, for 10 physicians. A relationship between the number of intermittent shots needed to complete a procedure and the patient DLP is demonstrated.
314
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
Fig. 1. Physician experience (years) vs. average Dose-Length Product (DLP) mGy-cm. Numbers next to each data point represent the years of experience for that physician.
Fig. 2. Linear regression analysis showing the number of intermittent CT shots required to complete a procedure vs. the patient DLP (mGy-cm). Physician labels and years of experience are also shown.
3.3. SmartStep use vs. Conventional CT When the use of Smartstep vs non-use of SmartStep was analyzed for the entire physician group, our statistic results show that there is a significant difference (p < 0.01) between the doses patients received (DLP = 539.8 ± 169.4 mGy-cm) when physicians used SmartStep vs. when Smartstep was not used (DLP = 1269.7 ± 659 mGy-cm).
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
315
Fig. 3. Average DLP per physician practice, showing the total number of procedures each physician performed. These numbers are for both procedures with and without Smartstep where relevant.
3.4. Influence of SmartStep experience When we take out the data for physicians who have never used SmartStep, and focus only on those physicians who have used SmartStep at some time in their practice, we find an interesting result. When Smartstep was used (n = 26), the average patient dose (DLP) was 539.8 ± 169.4 mGy-cm. When Smartstep was not used (n = 23), the average patient dose was 1133 ± 349 mGy-cm. This implies that for those physicians who used smartstep for some procedures and did not use it for others, there was a significant difference (p < 0.01) in their practice in regards to patient dose. For three physicians (I, G, and D) who used smartstep for some procedures and did not use it for others, the doses patients received were very comparable (Fig. 3). While physicians C & J performed procedures where the patient doses were on the high end leading to the high average for this group. 3.5. Patient Detriment Index (PDI) The average DLP (mGy-cm) was compared for each physician. The ratio of the total average DLP associated with each physician was normalized against the median DLP value for the group. As shown in Fig. 4, of the six physicians with a PDI < 1, four have used SmartStep. Of the four physicians with PDI > 1, one (C) has used the SmartStep feature. Of the 9 biopsies performed by physician C, 3 involved smartstep at significantly lower doses (DLP = 346 mGy-cm) than the 6 without smartstep (DLP = 2426
316
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
Fig. 4. Patient Detriment Index (PDI) values are shown above with the number of procedures each physician performed. The solid black line represents the PDI threshold value of 1. PDI less than one indicates good practice while PDI more than one is considered to be detrimental to the patient.
mGy-cm). Furthermore, all those physicians who used SmartStep, except C, have a lower relative PDI compared to their PDI without SmartStep. Three of five physicians who did not use SmartStep, have PDIs > 1 (Fig. 4). 3.6. Odds ratio calculation The odds ratio is 7.7 with a 95% confidence level and a range of 3.26 to 18.10. 4. Discussion Physician practice during CT guidance of biopsy plays a major role in determining patient radiation dose. When physicians E and H were compared for the use of conventional CT intermittent shots, a significant difference was found between the dose to patients and the number of intermittent shots taken. Both physicians did not use SmartStep during procedures for the time period analyzed. However, physician E had consistently lower doses than physician H. Doses were 720.4 ± 77.4 mGy-cm vs. 1558.2 ± 294.9 mGy-cm on the average for physicians E and H respectively for lung biopsy data analyzed over a six month period. These numbers can be partly explained by the fact that physician H required higher mA values (289 ± 50.9 mA) compared to physician E (210 ± 35 mA). In addition, physician E required fewer CT intermittent shots; 9.9 ± 2.5 s compared to physician H, who required 12.1 ± 3.7 s intermittent shots to complete lung biopsy studies during the six month period considered. The choice of a combination of more intermittent shots and higher mA requirements will lead to higher patient doses
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
317
for H compared to E. One plausible explanation of the difference between these physicians is their expectations of image quality. This is evidenced by the fact that the average patient dimensions for the both physicians were identical. For physician E, patient average diameter was 294.2 mm, and for physician H, average patient diameter was 296.3 mm. We would therefore expect that image noise for the two patient populations would not be far apart, so the use of more mA (289 ± 50.9 mA) by H compared to E (210 ± 35 mA) may be purely a preference for image appearance. We also expected that a procedure’s degree of difficulty may play a role in how many intermittent shots was required for completion. Upon analysis of various random procedures performed by both physicians, we found the performance of the two physicians to be consistent with the data in Table 1. The general trend for these procedures was the same for both E and H, with H requiring in general higher mA and longer times to complete procedures. This result is seen in Table 2 where a wide variety of procedures are shown, but with one general trend, and that is the difference in DLP and number of shots needed to complete the procedure. Differences between procedures with and without the use of SmartStep data were significant. In analysing the data for SmartStep use, our results revealed that when SmartStep was used by those physicians who used Smartstep some of the time, those physicians had high doses (DLP = 1133 mGy-cm) when they did not use SmartStep. These high doses were comparable to the doses patients received (DLP = 1269.75 mGy-cm) when they were treated without the use of SmartStep. One would expect that physicians who have used a dose savings feature would be more sensitive to reducing dose when they used intermittent shots than their peers who had never used a dose savings feature. This sensitivity to reducing patient dose would be reflected in the willingness to use a lower mA for intermittent shots when they did not use the dose savings feature. The fact that the data for the two groups is comparable would suggest that physicians need to be more proactive when it comes to scanning parameters used during their biopsy procedures. However, because of the small sample size, this average data could be misleading. For three Physicians I, G, D, the results were quite interesting. Physicians I, G, D used Smartstep for some procedures and did not use smartstep for other procedures. Their results show that the patient doses were not significantly different whether they used Smartstep or not. This is a desirable result showing that these physicians are mindful about using low dose techniques whether they used Smartstep or not. While this is encouraging, more data is needed to reach a solid conclusion about the practice for this group of physicians who used Smartstep for some procedures and not for others. It must be pointed out that physicians C & J performed procedures where the patient doses were on the high end leading to the high average for this group. We also considered the possibility that physician experience may play a role in the dose patients received when biopsies were performed. It is expected that a less experienced physician may use more radiation in order to have a better image quality for performing the biopsy. Figure 1 shows that there was no strong correlation between experience and the DLP values (R2 = 0.0226). Currently vendors provide dose savings features on their scanner but not all physician take advantage of these features. This leads to large variations in dose to patients. Our results show that this feature can play a large role in reducing dose to patients, and can potentially minimize large variations in patient dose caused by physician practice or preference. When comparing use vs non-use of SmartStep, it must be pointed out that the data was not collected on the same patient during a given procedure, but on different patients. While for different patients, the same procedure may be done over longer times and experience different complications, with a large enough dataset of patients, we expect such variations to be smoothed out. Nevertheless, we can still see a difference between doses and times for the use of dose savings or non-use of dose savings features. We compared the average total DLP (mGy-cm) for each physician with the median total DLP for the entire group over the one year period. The ratio of the total DLP associated with each physician was
318
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
normalized against the median value and called the patient detriment index (PDI). This index compares by how much each physician deviates from the entire group, and can serve as a way to rank physicians by their practice. Since the data was only approximately normally distributed, it was more appropriate to use the median values instead of the mean values. The mean in this situation will not represent the central tendency of the distribution. For data that has a Gaussian distribution, the average will be equal to the median, so using the average or the median will lead to the same result. In Fig. 3, It can be seen that although some physicians (E, A) did not use smartstep, their PDIs were still lower than 1. On the other hand, C used smartstep but had a PDI>1. C performed 9 studies during the time period analyzed. Of this number 3 were with Smartstep where the doses were low, and 6 without Smartstep. The 6 without Smartstep had the highest doses among the group of physicians. The utility of PDI is that it reveals the difference between physicians in the group. Since SmartStep saves dose, not using it may lead to a higher PDI compared to those who use SmartStep. Even if all physicians used SmartStep, PDIs will be a way to compare physicians with their peers in the group since there will be a lower median dose threshold to use as a benchmark, and all physicians will be compared to this lower threshold number. This will help drive doses even lower. In short, the use of a PDI could help rank physicians according to their practice. Healthcare organization could use these rankings to make improvements by defining their own patient detriment index. Also, the PDI can be applied to any health care center to allow physicians to minimize radiation dose by providing a metric or guidelines related to their practice in that institution relative to the standard of practice in the community. Ultimately, it could be used to compare practices at various institutions and could help to develop a national standard once more data from various institutions is available. With full implementation of the Affordable Care Act in 2014, the PDI could be included in patient care metrics to demonstrate how healthcare providers are working to provide the best possible care to patients. It can be seen how a concept like the PDI or similar ideas could be implemented in medical Imaging departments nationwide. Another potential benefit of using the PDI is to set a threshold dose for the purpose of Joint commision sentinel dose events monitoring in CT guided interventional procedures. Currently, with all the attention that fluroscopy is receiving, the time has come to set thresholds not only for CT guided interventions but also for CT procedures in general. The odds ratio was used to indicate the likelihood of patients receiving a DLP above the median DLP threshold value or below the threshold value. The odds ratio for a DLP above the threshold value was 7.7. Since this number was larger than 1, we concluded that patients in the “treatment” group (group treated by physician H), were more likely than patients in the “control” group (treated by physician E) to receive a DLP above the threshold value. The odds ratio is calculated with a 95% confidence interval, and a range from 3.23 to 18.10. In other words, patients treated by H have their risks of receiving a dose above the threshold value increased by 7.7 times, with a range of 3.26 to 18.1, as compared to those treated by E. Lastly, this study is very important to provide information to staff, and to provide guidelines on training staff to minimize radiation dose to patients during CT guided interventional procedures. Potential guidelines include, using dose savings features available for biopsies, using lower mA values for intermittent CT shots, and also using the fewest number of CT shots possible to complete a procedure. A limitation of this study could be the wide variation between physicians on the number and types of procedures performed. However, we believe this limitation is not a detriment to our study, it being a study that can provide practice guideline for physicians and staff on ways to reduce patient doses. Another limitation is the limited data set of procedures performed by some physicians. A larger multi-hospital
V.J. Weir et al. / Impact of physician practice on patient radiation dose during CT guided biopsy procedures
319
study involving a larger number of physicians over a few years may be necessary to set a standard of practice for a specific procedure. For the odds ratio calculation, the assignment of a “treatment” group, as patients treated by physician H, or a “control” group, as patients treated by physician E can be questioned. We believe our application provides a valid approximation since we have set a threshold value for DLP that we expect can be used a a benchmark for “treatment”: when patients receive a dose above this threshold, vs “control”, when patients receive doses below this threshold value. To our knowledge, this study presents for the first time the concept of the PDI as a way of comparing the practice of physicians with their peers. We also present for the first time the use of the odds ratio in the context of medical practice. 5. Conclusion Scanner reported patient doses can be significantly influenced by physician practice during CT guided biopsy procedures. The potential impact may be mitigated by the use of the following strategies. First, a dedicated biopsy protocol should be created, based on right practice, right dose and best outcome as outlined by the American College of Radiology. Second, physicians should use dose savings features such as SmartStep when available. Third, if intermittent shots are used, lower mA values and fewer intermittent shots should be used. With the current push to record patient doses on interventional procedures, setting up a local dose threshold based on physician practice in an institution can help healthcare providers identify areas where improvements need to be made with regards to lowering patient dose. Such improvements can be made by combinations of practice guidelines, staff training, equipment upgrades, and incentives, among others. References [1] [2] [3] [4] [5] [6] [7] [8] [9]
S. Sheth, U.M. Hamper, D.B. Stanley et al., US Guidance for Thoracic Biopsy: A Valuable Alternative to CT, Radiology 210 (1999), 726–726. J. Stattaus, S. Maderwald, M. Forsting et al., MR guided core biopsy with MR fluoroscopy using a short, wide-bore 1.5-Tesla scanner: feasibility and initial results, J Magn Reson Imaging 27 (2008), 1181–1187. J.R. Haaga and R.J. Alfidi, Precise biopsy localization by computer tomography, Radiology 118 (1976), 603–607. K. Katada, H. Anno, Y. Ogura et al., Development and early trials of real-time CT fluoroscopy. Neuroradiology 37 (1995), 587–588. S. Silverman, K. Tuncali, D. Adams et al., CT Fluoroscopy-guided Abdominal Interventions: Techniques, Results, and Radiation Exposure, Radiology 212 (1999), 673–681. V. Tsapaki, C. Triantopoulou, P. Maniatis et al., Patient Skin Dose assessment during CT-guided interventional procedures, Radiation Protection Dosimetry 129(1–3) (2008), 29–31. E. Paulson, D. Sheafor, D. Enterline et al., CT Fluoroscopy – guided Interventional Procedures: Techniques and Radiation Dose to Radiologists, Radiology 220 (2001), 1612–1617. B. Daly, T.L. Krebs, J.J. Wong-You-Cheong et al., Percutaneous Abdominal and Pelvic Interventional Procedures Using CT Fluoroscopy Guidance, AJR 173 (1999), 637–644. W. Teeuwisse, J. Geleijns, J. Broerse et al., Patient and staff dose during CT guided biopsy, drainage and coagulation, The British Journal of Radiology 74 (2001), 720–726.
Copyright of Journal of X-Ray Science & Technology is the property of IOS Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
Copyright of Journal of X-Ray Science & Technology is the property of IOS Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
