The Journal of Emergency Medicine, Vol. -, No. -, pp. 1–9, 2014 Copyright Ó 2014 Elsevier Inc. Printed in the USA. All rights reserved 0736-4679/$ - see front matter
http://dx.doi.org/10.1016/j.jemermed.2014.01.030
Selected Topics: Emergency Radiology
RADIATION EXPOSURE FROM IMAGING TESTS IN PEDIATRIC EMERGENCY MEDICINE: A SURVEY OF PHYSICIAN KNOWLEDGE AND RISK DISCLOSURE PRACTICES Kathy Boutis, MD,* Jason Fischer, MD,* Stephen B. Freedman, MD,† and Karen E. Thomas, MD‡ *Division of Emergency Medicine, Department of Pediatrics, The Hospital for Sick Children and University of Toronto, Toronto, Ontario, Canada, †Sections of Emergency Medicine and Gastroenterology, Department of Pediatrics, Alberta Children’s Hospital and Alberta Children’s Hospital Research Institute, University of Calgary, Calgary, Alberta, Canada, and ‡Department of Diagnostic Imaging, The Hospital for Sick Children, and University of Toronto, Toronto, Ontario, Canada Reprint Address: Kathy Boutis, MD, The Hospital for Sick Children, 555 University Avenue, Toronto, ON M5G 1X8, Canada
, Abstract—Background: Disclosing potential future malignancy risks from diagnostic tests that expose children to ionizing radiation in the emergency department may be challenging. Objectives: We determined the proportion of pediatric emergency medicine (PEM) physicians who are aware of current malignancy risk estimates associated with head computed tomography (CT). We also examined reported risk and strategy disclosure practice patterns. Methods: We conducted an online survey of members of a national Canadian PEM physician association using a modified Dillman’s technique. Results: Of 156 eligible participants, 126 (80.8%) responded to the survey. Of the 126 respondents, 124 (98.4%; 95% confidence interval [CI] 96.2–100) reported that there is a potential malignancy risk associated with head CT, and 46 (36.5%; 95% CI 28.1–44.9) correctly identified the best current estimate of this risk. The majority, 68.8% (95% CI 60.7–76.9), reported disclosing these possible risks ‘‘most of the time/almost always.’’ Although some physicians reported varying their strategy with the clinical scenario, the most frequently selected disclosure strategies were a comparison with chest radiographs and everyday risks. Frequently cited barriers to informed risk-benefit discussions were concerns that parents will worry excessively about cancer (27.8%), discussions during the treatment of a critically ill child (23.8%), and a concern that parents may not want the test (15.9%). Conclusions: Approximately one-third of pediatric emergency physicians were able to identify the best available
estimate of the malignancy risk from a head CT. Although there are some barriers, many PEM physicians report regularly participating in risk-benefit disclosures. Ó 2014 Elsevier Inc. , Keywords—child; radiation; computed tomography; management
INTRODUCTION Numbers of annual computed tomography (CT) examinations have been increasing incrementally in most countries by about 10% each year during the last 10–20 years (1). In particular, CT has become a vital component in the urgent diagnostic evaluation of pediatric emergency department (ED) patients, with usage having increased fivefold in recent years (2). However, there are increasing concerns about the potential future malignancy risk associated with exposure to ionizing radiation in children (3,4). Concerns are heightened for CT, which often involves exposure to higher radiation doses than other diagnostic imaging modalities (5). However, disclosing these potential risks to parents in an ED may be challenging, as the amount of time available to engage in an informed discussion may be limited. In addition, emergency physicians may have concerns that this could
RECEIVED: 3 August 2013; FINAL SUBMISSION RECEIVED: 31 October 2013; ACCEPTED: 31 January 2014 1
2
K. Boutis et al.
lead to parental resistance to proceed with a CT scan that is clinically indicated. Several studies to date have examined physician knowledge of radiation risks in a variety of practice settings. Most have involved physicians caring for adult patients, with a small number of studies directed towards physicians caring for pediatric patients (6–16). Regardless of setting, all have raised concern about deficiencies in physician knowledge regarding the radiation doses associated with diagnostic tests and their associated malignancy risks. No studies have specifically evaluated pediatric emergency medicine (PEM) physicians, a specialty where the relevance of this information to daily practice is particularly high. In addition, we are unaware of any published studies examining strategies employed by physicians to disclose these risks and the perceived barriers faced in initiating such discussions. We performed a survey of Canadian PEM physicians to determine the proportion who are aware of the most accurate current malignancy risk estimate associated with head CT. In addition, we examined risk disclosure practice patterns and the specific strategies used by these physicians in the ED during discussions with families and patients. MATERIALS AND METHODS Survey Design and Study Population Between August and October 2012, a self-administered online survey was sent to all attending-level physicians who were members of Pediatric Emergency Research Canada (PERC), a national organization that houses a database of PEM physicians. Approximately 75% of all PEM physicians are members of PERC. There are PEM physicians in all of the Canadian provinces, but none in the Territories. Members who did not receive the survey due to technical reasons, did not have an accurate electronic e-mail address on file, or were not staff-/ attending-level physicians in Canada working primarily in an ED, were excluded. The study was approved by the institutional research ethics board and PERC prior to administration. Finally, the survey was constructed and reported in accordance with the published recommendations of the Journal of Medical Internet Research (17). Survey Content A literature search did not reveal any validated questionnaire for our survey content, and thus we developed a survey in accordance with the methods advocated by Streiner and Norman as well as Burns et al. (18,19). The questions and estimates of radiation dose and respective potential risks were based on the best available information from
the relevant literature, three PEM physicians with survey expertise, and one pediatric radiologist with survey and content expertise (3–5,20). Radiation dose information used to derive equivalent time periods of background radiation was based on 2011 institutional effective dose estimates of 0.02 mSv for a two-view chest radiograph on a 5-year-old child, 1.5–2.0 mSv for a single-phase (noncontrast) pediatric head CT scan, and an annual background radiation exposure of 3 mSv (5,21). Potential future excess malignancy risk estimates were based on BEIR VII data, and those available from Image Gently, the international pediatric radiation safety awareness campaign (3,20). Although the ‘‘most correct estimate’’ of malignancy risk estimates used were approximately one in a million for a chest radiograph series and one in 10,000 for head CT, the authors acknowledge that risk will vary according to age and gender of a child. Items for the survey were generated by an expert panel until no new items emerged, distributed over five consensus-based sections. The items were then pre- and pilot-tested on 20 PEM physicians based outside of Canada. Initially, we provided the survey on paper to 10 of these PEM physicians, and we asked for specific feedback. The revised questions were loaded onto the electronic survey platform and pilot-tested on 10 additional PEM physicians distinct from previous PEM participants. These PEM physicians provided input on survey flow, userfriendliness, question clarity/content, and time to completion, and survey questions were removed or modified in accordance with feedback. The final survey included 20 questions, and limited data collection time to 5–10 min per participant. The final survey (Supplemental document S1) addressed the following domains (with respective number of questions): demographics (6 questions), knowledge of potential risks associated with ionizing radiation in imaging (5 questions), risk disclosure (6 questions), and public awareness of radiation exposure from diagnostic imaging tests (3 questions). After completion of the survey, physicians were provided the opportunity to review a Webinar on radiation exposure and potential malignancy risk associated with frequently ordered diagnostic imaging tests and approaches to informing patients about the small potential malignancy risks. Survey Administration The survey was administered using www.surveygizmo. com, which allowed for an unlimited number of questions and responses, data collection via Web-link and e-mail, forced responses for each question, the use of skip logic, a progress bar, and downloading to a spreadsheet. Multiple-item screens to decrease completion time and minimize incomplete responses were incorporated into the survey design.
PEM Physician Knowledge of Radiation Risks
The survey was sent out to potential respondents using a modified Dillman’s tailored design method for mail and Internet surveys (22). This approach included an initial e-mail with an introductory letter and a link to the survey. Links were unique to each participant and cookies were used to assign a unique identifier to each participant computer, thereby eliminating the potential for multiple responses being provided by the same individual. Four reminder e-mails, which included corresponding letters and accrual rates, were also sent at 3-week intervals. Each reminder included details about the investigators, the purpose and duration of the survey, and assured confidentiality. De-identified demographic variables on nonresponders were available from the PERC member database. Outcomes Primary outcome was the proportion of PEM physicians who identified the correct potential malignancy risk estimate associated with head CT in children. Secondary outcomes included the proportion of PEM physicians who were aware of any potential malignancy risk from CT; PEM physician demographics (years since certification, practice setting, certification type) associated with an awareness of most correct risk estimate; proportion of response options for radiation dose exposure from chest radiographs and CT; proportion of responses of PEM physicians who disclosed risks reported as either ‘‘almost always, most of the time, sometimes, infrequently, almost never’’; proportion of response options for barriers and strategies used for risk disclosure in the ED; and proportion of responses for all applicable sources of knowledge of radiation risks and exposures. Data Analyses There are approximately 185 physician members of the PERC database, and we anticipated a response rate of approximately 75% (23). This final sample size of 140 from a population of 250 (estimated number of Canadian PEM physicians) produces a 95% confidence interval (CI) around the sample proportion of 6 5% when the estimated proportion of PEM physicians who correctly estimate potential cancer risks from a head CT is 25%. Descriptive statistics were used to summarize responses. Proportions were compared using the chi-squared test. Logistic regression was used to determine association between PEM physician demographic variables and the primary outcome. All variables were entered into a full (i.e., saturated) multi-variable logistic regression model. Goodness-of-fit of final model to the data was tested using the Hosmer-Lemeshow test. We compared demographic variables between responders and nonresponders. All ana-
3
lyses were completed using SPSS for Windows (version 13; SPSS Inc., Chicago, IL). RESULTS Enrollment and Demographics Of the 187 PEM physicians listed in the PERC database, 156 were eligible for participation (28 were not attending PEM physicians, 3 were involved in generation of the survey questions) and 126 (80.8%) responded to the survey. All respondents completed all survey questions. Of the participants, 55.6% were certified in PEM, 73.8% were full time, and 32.5% had worked for 6–10 years at a university-affiliated children’s hospital (Table 1). In addition, there was representation of PEM physicians from each province (Figure 1). On completion of the survey, 107 participants (84.9%) downloaded the optional Webinar. There were no significant differences between responders and nonresponders in years since completion of training (p = 0.11), primary practice setting (p = 0.24), responder province (p = 0.32), or highest level of certification (p = 0.12). PEM Physician Knowledge of Potential Malignancy Risk and Radiation Dose (Table 2) Of the 126 physician respondents, the percentage (95% CI) who believed there was an increased malignancy risk after head CT was 98.4% (95% CI 96.2–100), and 36.5% (95% CI 28.1–44.9) accurately reported the best
Table 1. Demographics of Physician Participants Variable Number of years in practice since completion of training, n (%) #5 6–10 11–15 16–20 >20 Full time in pediatric emergency medicine, n (%) Certification, n (%) Pediatrics with pediatric emergency medicine Pediatrics Other (emergency and family medicine) Practice setting, n (%) University-affiliated children’s hospital University-affiliated general hospital
Pediatric Emergency Physicians n = 126
25 (19.8) 41 (32.5) 19 (15.1) 20 (15.9) 21 (16.7) 93 (73.8) 70 (55.6) 34 (27.0) 22 (17.5) 114 (90.5) 12 (9.5)
4
K. Boutis et al.
Figure 1. Participating pediatric emergency physicians from each province in Canada.
current estimate of this risk (primary outcome) at 1 in 10,000. The risk was underestimated by 27.8% (95% CI 20.0–35.6) and overestimated by 34.1% (95% CI 25.8– 42.4). There was no statistical association between knowledge of the most correct cancer risk estimate and any demographic variable (years since certification, p = 0.3; practice setting, p = 0.4; or certification type, p = 0.6; Hosmer-Lemeshow p = 0.7; area under the curve 0.7 [95% CI 0.61–0.8]). A similar percentage of participants, 96.8% (95% CI 93.7–99.9) were aware that there is also a potential malignancy risk associated with radiographs such as chest radiographs, with 77.8% (95% CI 70.1–85.1) choosing the correct, much lower, associated risk of < 1 in 1,000,000. Table 2. Pediatric Emergency Physician Knowledge of Potential Malignancy Risk from Imaging Tests that Use Ionizing Radiation
Almost no risk at all (
http://dx.doi.org/10.1016/j.jemermed.2014.01.030
Selected Topics: Emergency Radiology
RADIATION EXPOSURE FROM IMAGING TESTS IN PEDIATRIC EMERGENCY MEDICINE: A SURVEY OF PHYSICIAN KNOWLEDGE AND RISK DISCLOSURE PRACTICES Kathy Boutis, MD,* Jason Fischer, MD,* Stephen B. Freedman, MD,† and Karen E. Thomas, MD‡ *Division of Emergency Medicine, Department of Pediatrics, The Hospital for Sick Children and University of Toronto, Toronto, Ontario, Canada, †Sections of Emergency Medicine and Gastroenterology, Department of Pediatrics, Alberta Children’s Hospital and Alberta Children’s Hospital Research Institute, University of Calgary, Calgary, Alberta, Canada, and ‡Department of Diagnostic Imaging, The Hospital for Sick Children, and University of Toronto, Toronto, Ontario, Canada Reprint Address: Kathy Boutis, MD, The Hospital for Sick Children, 555 University Avenue, Toronto, ON M5G 1X8, Canada
, Abstract—Background: Disclosing potential future malignancy risks from diagnostic tests that expose children to ionizing radiation in the emergency department may be challenging. Objectives: We determined the proportion of pediatric emergency medicine (PEM) physicians who are aware of current malignancy risk estimates associated with head computed tomography (CT). We also examined reported risk and strategy disclosure practice patterns. Methods: We conducted an online survey of members of a national Canadian PEM physician association using a modified Dillman’s technique. Results: Of 156 eligible participants, 126 (80.8%) responded to the survey. Of the 126 respondents, 124 (98.4%; 95% confidence interval [CI] 96.2–100) reported that there is a potential malignancy risk associated with head CT, and 46 (36.5%; 95% CI 28.1–44.9) correctly identified the best current estimate of this risk. The majority, 68.8% (95% CI 60.7–76.9), reported disclosing these possible risks ‘‘most of the time/almost always.’’ Although some physicians reported varying their strategy with the clinical scenario, the most frequently selected disclosure strategies were a comparison with chest radiographs and everyday risks. Frequently cited barriers to informed risk-benefit discussions were concerns that parents will worry excessively about cancer (27.8%), discussions during the treatment of a critically ill child (23.8%), and a concern that parents may not want the test (15.9%). Conclusions: Approximately one-third of pediatric emergency physicians were able to identify the best available
estimate of the malignancy risk from a head CT. Although there are some barriers, many PEM physicians report regularly participating in risk-benefit disclosures. Ó 2014 Elsevier Inc. , Keywords—child; radiation; computed tomography; management
INTRODUCTION Numbers of annual computed tomography (CT) examinations have been increasing incrementally in most countries by about 10% each year during the last 10–20 years (1). In particular, CT has become a vital component in the urgent diagnostic evaluation of pediatric emergency department (ED) patients, with usage having increased fivefold in recent years (2). However, there are increasing concerns about the potential future malignancy risk associated with exposure to ionizing radiation in children (3,4). Concerns are heightened for CT, which often involves exposure to higher radiation doses than other diagnostic imaging modalities (5). However, disclosing these potential risks to parents in an ED may be challenging, as the amount of time available to engage in an informed discussion may be limited. In addition, emergency physicians may have concerns that this could
RECEIVED: 3 August 2013; FINAL SUBMISSION RECEIVED: 31 October 2013; ACCEPTED: 31 January 2014 1
2
K. Boutis et al.
lead to parental resistance to proceed with a CT scan that is clinically indicated. Several studies to date have examined physician knowledge of radiation risks in a variety of practice settings. Most have involved physicians caring for adult patients, with a small number of studies directed towards physicians caring for pediatric patients (6–16). Regardless of setting, all have raised concern about deficiencies in physician knowledge regarding the radiation doses associated with diagnostic tests and their associated malignancy risks. No studies have specifically evaluated pediatric emergency medicine (PEM) physicians, a specialty where the relevance of this information to daily practice is particularly high. In addition, we are unaware of any published studies examining strategies employed by physicians to disclose these risks and the perceived barriers faced in initiating such discussions. We performed a survey of Canadian PEM physicians to determine the proportion who are aware of the most accurate current malignancy risk estimate associated with head CT. In addition, we examined risk disclosure practice patterns and the specific strategies used by these physicians in the ED during discussions with families and patients. MATERIALS AND METHODS Survey Design and Study Population Between August and October 2012, a self-administered online survey was sent to all attending-level physicians who were members of Pediatric Emergency Research Canada (PERC), a national organization that houses a database of PEM physicians. Approximately 75% of all PEM physicians are members of PERC. There are PEM physicians in all of the Canadian provinces, but none in the Territories. Members who did not receive the survey due to technical reasons, did not have an accurate electronic e-mail address on file, or were not staff-/ attending-level physicians in Canada working primarily in an ED, were excluded. The study was approved by the institutional research ethics board and PERC prior to administration. Finally, the survey was constructed and reported in accordance with the published recommendations of the Journal of Medical Internet Research (17). Survey Content A literature search did not reveal any validated questionnaire for our survey content, and thus we developed a survey in accordance with the methods advocated by Streiner and Norman as well as Burns et al. (18,19). The questions and estimates of radiation dose and respective potential risks were based on the best available information from
the relevant literature, three PEM physicians with survey expertise, and one pediatric radiologist with survey and content expertise (3–5,20). Radiation dose information used to derive equivalent time periods of background radiation was based on 2011 institutional effective dose estimates of 0.02 mSv for a two-view chest radiograph on a 5-year-old child, 1.5–2.0 mSv for a single-phase (noncontrast) pediatric head CT scan, and an annual background radiation exposure of 3 mSv (5,21). Potential future excess malignancy risk estimates were based on BEIR VII data, and those available from Image Gently, the international pediatric radiation safety awareness campaign (3,20). Although the ‘‘most correct estimate’’ of malignancy risk estimates used were approximately one in a million for a chest radiograph series and one in 10,000 for head CT, the authors acknowledge that risk will vary according to age and gender of a child. Items for the survey were generated by an expert panel until no new items emerged, distributed over five consensus-based sections. The items were then pre- and pilot-tested on 20 PEM physicians based outside of Canada. Initially, we provided the survey on paper to 10 of these PEM physicians, and we asked for specific feedback. The revised questions were loaded onto the electronic survey platform and pilot-tested on 10 additional PEM physicians distinct from previous PEM participants. These PEM physicians provided input on survey flow, userfriendliness, question clarity/content, and time to completion, and survey questions were removed or modified in accordance with feedback. The final survey included 20 questions, and limited data collection time to 5–10 min per participant. The final survey (Supplemental document S1) addressed the following domains (with respective number of questions): demographics (6 questions), knowledge of potential risks associated with ionizing radiation in imaging (5 questions), risk disclosure (6 questions), and public awareness of radiation exposure from diagnostic imaging tests (3 questions). After completion of the survey, physicians were provided the opportunity to review a Webinar on radiation exposure and potential malignancy risk associated with frequently ordered diagnostic imaging tests and approaches to informing patients about the small potential malignancy risks. Survey Administration The survey was administered using www.surveygizmo. com, which allowed for an unlimited number of questions and responses, data collection via Web-link and e-mail, forced responses for each question, the use of skip logic, a progress bar, and downloading to a spreadsheet. Multiple-item screens to decrease completion time and minimize incomplete responses were incorporated into the survey design.
PEM Physician Knowledge of Radiation Risks
The survey was sent out to potential respondents using a modified Dillman’s tailored design method for mail and Internet surveys (22). This approach included an initial e-mail with an introductory letter and a link to the survey. Links were unique to each participant and cookies were used to assign a unique identifier to each participant computer, thereby eliminating the potential for multiple responses being provided by the same individual. Four reminder e-mails, which included corresponding letters and accrual rates, were also sent at 3-week intervals. Each reminder included details about the investigators, the purpose and duration of the survey, and assured confidentiality. De-identified demographic variables on nonresponders were available from the PERC member database. Outcomes Primary outcome was the proportion of PEM physicians who identified the correct potential malignancy risk estimate associated with head CT in children. Secondary outcomes included the proportion of PEM physicians who were aware of any potential malignancy risk from CT; PEM physician demographics (years since certification, practice setting, certification type) associated with an awareness of most correct risk estimate; proportion of response options for radiation dose exposure from chest radiographs and CT; proportion of responses of PEM physicians who disclosed risks reported as either ‘‘almost always, most of the time, sometimes, infrequently, almost never’’; proportion of response options for barriers and strategies used for risk disclosure in the ED; and proportion of responses for all applicable sources of knowledge of radiation risks and exposures. Data Analyses There are approximately 185 physician members of the PERC database, and we anticipated a response rate of approximately 75% (23). This final sample size of 140 from a population of 250 (estimated number of Canadian PEM physicians) produces a 95% confidence interval (CI) around the sample proportion of 6 5% when the estimated proportion of PEM physicians who correctly estimate potential cancer risks from a head CT is 25%. Descriptive statistics were used to summarize responses. Proportions were compared using the chi-squared test. Logistic regression was used to determine association between PEM physician demographic variables and the primary outcome. All variables were entered into a full (i.e., saturated) multi-variable logistic regression model. Goodness-of-fit of final model to the data was tested using the Hosmer-Lemeshow test. We compared demographic variables between responders and nonresponders. All ana-
3
lyses were completed using SPSS for Windows (version 13; SPSS Inc., Chicago, IL). RESULTS Enrollment and Demographics Of the 187 PEM physicians listed in the PERC database, 156 were eligible for participation (28 were not attending PEM physicians, 3 were involved in generation of the survey questions) and 126 (80.8%) responded to the survey. All respondents completed all survey questions. Of the participants, 55.6% were certified in PEM, 73.8% were full time, and 32.5% had worked for 6–10 years at a university-affiliated children’s hospital (Table 1). In addition, there was representation of PEM physicians from each province (Figure 1). On completion of the survey, 107 participants (84.9%) downloaded the optional Webinar. There were no significant differences between responders and nonresponders in years since completion of training (p = 0.11), primary practice setting (p = 0.24), responder province (p = 0.32), or highest level of certification (p = 0.12). PEM Physician Knowledge of Potential Malignancy Risk and Radiation Dose (Table 2) Of the 126 physician respondents, the percentage (95% CI) who believed there was an increased malignancy risk after head CT was 98.4% (95% CI 96.2–100), and 36.5% (95% CI 28.1–44.9) accurately reported the best
Table 1. Demographics of Physician Participants Variable Number of years in practice since completion of training, n (%) #5 6–10 11–15 16–20 >20 Full time in pediatric emergency medicine, n (%) Certification, n (%) Pediatrics with pediatric emergency medicine Pediatrics Other (emergency and family medicine) Practice setting, n (%) University-affiliated children’s hospital University-affiliated general hospital
Pediatric Emergency Physicians n = 126
25 (19.8) 41 (32.5) 19 (15.1) 20 (15.9) 21 (16.7) 93 (73.8) 70 (55.6) 34 (27.0) 22 (17.5) 114 (90.5) 12 (9.5)
4
K. Boutis et al.
Figure 1. Participating pediatric emergency physicians from each province in Canada.
current estimate of this risk (primary outcome) at 1 in 10,000. The risk was underestimated by 27.8% (95% CI 20.0–35.6) and overestimated by 34.1% (95% CI 25.8– 42.4). There was no statistical association between knowledge of the most correct cancer risk estimate and any demographic variable (years since certification, p = 0.3; practice setting, p = 0.4; or certification type, p = 0.6; Hosmer-Lemeshow p = 0.7; area under the curve 0.7 [95% CI 0.61–0.8]). A similar percentage of participants, 96.8% (95% CI 93.7–99.9) were aware that there is also a potential malignancy risk associated with radiographs such as chest radiographs, with 77.8% (95% CI 70.1–85.1) choosing the correct, much lower, associated risk of < 1 in 1,000,000. Table 2. Pediatric Emergency Physician Knowledge of Potential Malignancy Risk from Imaging Tests that Use Ionizing Radiation
Almost no risk at all (
