Baseline Survey of the Neuroradiology Work Environment in the United States with Reported Trends in Clinical Work, Nonclinical Work, Perceptions of Trainees, and Burnout Metrics

A voluntary survey was sent to practicing and out-of-training members of the ASNR. Four hundred thirty-two respondents across a broad range of experience reported: 93% with workdays extending at least 1 hour past expected; 71.9% reading more cases per hour compared with previous years; 79.5% sometimes-to-always interpreting cases faster than comfortable for optimal interpretation; and 67.8% sometimes or more often with inadequate time to discuss abnormal results. Burnout symptoms ranged between 49% and 75%. The trends and the correlations should be concerning to the leaders of radiology and warrant further monitoring. BACKGROUND AND PURPOSE: Neuroradiologists have faced continuously increasing clinical workloads. Our aim was to establish and report a baseline survey of the current neuroradiology work environment in the United States and of experiential changes in recent years. MATERIALS AND METHODS: A voluntary survey was sent to practicing and out-of-training members of the American Society of Neuroradiology in the United States. Selected measures included workday volume and length, burnout symptoms, participation in academic and practice-building duties; effects on perceived interpretation quality, communication of abnormal results, and consideration of early retirement or career changes, among others. RESULTS: Four hundred thirty-two respondents across a broad range of experience reported the following: 52.8% (224/424) with teaching responsibilities; 93% (399/430) with workdays extending at least 1 hour past expected, in 45% (193/430) frequently or always; 71.9% (309/430) reading more cases per hour compared to previous years; 79.5% (341/429) sometimes-to-always interpreting cases faster than comfortable for optimal interpretation; and 67.8% (292/431) sometimes or more often with inadequate time to discuss abnormal results. Burnout symptoms ranged between 49% and 75% (211/428 to 322/428) across 4 indices. For academic activities of teaching, mentoring, and research/publications, a mean of 94.3% reported cut-backs during the past few years. For practice-building activities, 92% reported cut-backs, 51.6% (222/429) considered early retirement, and 38.8% (167/429) considered changing careers. CONCLUSIONS: Increasing clinical demands have coincided with destructive effects in the work environment and the ability and desire of neuroradiologists in the United States to perform academic or practice-building duties with a substantial incidence of burnout symptoms. While this survey does not prove causation, the trends and the correlations should be concerning to the leaders of radiology and warrant further monitoring.

I n the United States, the work volume of neuroradiologists has mirrored that of radiology and health care in general, with increases in annual work productivity (number of Relative Value Units [RVUs]) during the past decade. 1 This is part of a longer term trend in increased workloads that occurred in the prior de-cade. 2 The rise in productivity demand has resulted in some practices requiring the use of specific work-output RVU targets as thresholds for compensation or hiring without adequate regard for their effects on workers or potential limits to human task performance. Neuroradiologists' target and total RVUs are the highest of radiology subspecialties, [3][4][5] and neuroradiologists may serve as a bellwether for the broader radiologist population. To date, there has not been a single national survey collecting data on the neuroradiologists' practice work environment evaluating compensation, workplace hiring practices, effects on productivity, and output as well as burnout, to our knowledge.
The emphasis on ever-increasing RVU targets may have downstream effects on radiology, such as increased speed of interpretation, decreased remaining time for training residents and fellows, decreased research output, decreased time for practice building (such as quality projects, administrative committees), and potential burnout of radiologists. Worker burnout is a known contributor to workplace errors and is costly in terms of personal and institutional impact. Burnout of radiologists and other physicians is becoming an important topic, not just in the United States, but also in other countries. [6][7][8] This survey of neuroradiologists was performed to establish a national baseline for work patterns in the United States to allow tracking and comparison with future surveys. The survey will be repeated and updated biennially to allow tracking and comparisons as the work environment in the United States continues to change. This schedule will provide better data to practitioners and practice managers for understanding, measuring, and (perhaps) modifying the work environment.

MATERIALS AND METHODS
A voluntary on-line survey was sent to practicing neuroradiologists in the United States who are members of the American Society of Neuroradiology (ASNR). It was inclusive of the neuroradiologists in the United States and excludes neuroradiologist members who were outside the United States. The survey was launched at the beginning of August 2016 and was open until October 5, 2016. The current membership includes practicing physicians and fellows in training. We excluded the latter because many of the questions interrogated trends and changes with a baseline that was longer than the training period. Although participation was voluntary, an incentive was offered in the form of a drawing for a gift certificate valued at $250, given the length of the survey.
The survey (On-line Appendix) was collected electronically by using Survey Monkey (http://www.surveymonkey.com), a survey Web site that allows viewing of survey responses only by the owner of the survey. The American Society of Neuroradiology agreed to send the survey out to members in the United States by e-mail. Members who chose to participate did so by clicking the e-mail link to open the Survey Monkey Web page containing the survey and completed it on their device. No direct contact occurred between the study Principal Investigator or the survey respondents except for announcement to the respondents of the winner of the prize associated with the lottery.
The e-mail addresses and surveys were initially linked temporarily only if the participant chose to participate in the prize lottery. The e-mail address was collected only for the lottery and then was separated from the survey response. The survey responses were then de-linked from any e-mail address in the investigator's response data base. Until the e-mail addresses were separated from the responses, they were considered confidential. Once the responses were separated from the e-mail addresses, there was no mechanism for tracking responses back to a specific e-mail address.
Survey data were then checked for quality and to remove duplicates. Internet Protocol addresses, survey submission time, and e-mails were cross-referenced to identify responses that could be considered duplicates.
In 1 case, duplicate answers from the same ASNR member were reconciled before inclusion in the analysis. Initial data display was performed by using the on-line survey tool output; then, more advanced data analysis was performed by importing the data into an Excel (Microsoft, Redmond, Washington) spreadsheet. Statistical comparison of selected yes/no results between groups was performed by using n-1 comparison.
Results were broken into several groups for analysis: overall results, respondents who routinely teach trainees versus respondents who do not routinely teach trainees, and roughly by demographic generation with practice experience as a proxy. If one assumed contiguous education and training with completion of the fellowship at 32 years of age, dividing respondents at and below 20 years of experience would roughly divide respondents into 2 major demographic generations. This division would approximately divide respondents at a date of birth of 1964, roughly "Generation X" including 1 or 2 years of "Millennials" versus "Baby Boomers" and "Greatest Generation" with Ͼ20 years of experience.

RESULTS
Article length constraints limit text reporting of the results to the overall results and selected subgroups. The results tables (Tables  1-3 and On-line Tables 1-3) provide more detailed comparisons of subgroups, including teaching versus nonteaching, and Ͻ20 years of experience versus Ͼ20 years of experience.
Only 1 duplicate response from 1 Internet Protocol address was discovered. Once the duplicated responses were reconciled and the 20 member-in-training responses were excluded, a total of 432 participants were included for further analysis. There was minimal survey fatigue; although questions could be skipped, 99.3% (429/432) of the respondents answered the last question. The results below are based on the percentage of participants who responded to a question.

Demographic Results
A slight majority of respondents, 52.8% versus 47.2%, reported that they routinely teach radiology residents or fellows, giving the survey a relatively well-balanced sample of both training settings and nontraining sites of neuroradiology work.
The dataset was also fairly well-distributed by experience from the first 5 years out of training to those with Ͼ31 years of posttraining work experience (Fig 1).
The survey group encompassed a fairly wide spread of respondents with regard to the anticipated number of years remaining in their careers. Fourteen percent of the respondents were within Յ5 years of planned retirement, with a peak of 20% planning to work 11-15 years and 19.5% planning to work Ն26 years. There was less diversity with regard to the number of jobs held: Three of 4 of the respondents (77.8%) had not changed employers in past few years, while 17.4 had changed employers once, 3.9% had changed twice, and Ͻ1% had changed Ն3 times.

Reported Compensation and Time Off
Slightly less than half, 48.5%, reported that their compensation versus the prior year was unchanged; 28.4% reported an increase; and 23.1% reported a decrease. Time off the clinical schedule (vacation and nonclinical professional time) was more stable, with 73% reporting no change since the prior year, but with a more concerning trend in that those reporting a decrease (18.7%) outnumbered those with an increase (8.4%). Similarly, if we looked at weeks rather than days, 70.8% reported no change, but those reporting working more weeks (23.2%) outnumbered those reporting working fewer (6.0%).

Perceptions of Clinical Workload and Time at Work
Close to one-third (32.5%) of respondents said that radiology compensation was determined by RVU targets. At a personal level, most (62.7%) reported that during the past few years, their RVU production had increased, while 6.5% reported a decrease, 16.4% reported that it was the same, and 14.4% did not know. Reporting on the trend of the number of hours worked per day during the past few years, almost half (49.7%) reported working longer days, while a slightly smaller number, 46.9%, reported that they worked about the same,   and only 3.5% reported a shorter workday. In a related question, the participants were asked how often they had to stay Ն1 hour beyond the expected workday. Here only slightly more than one-quarter reported that a longer stay never or rarely occurred (27.4%, sum of 2 responses). More than half the respondents reported that it happened sometimes or frequently (57.2%, sum of 2 responses), and most interesting were the 15.4% who said they always stayed an hour or more past the end of the "expected workday." With the regard to the number of workdays in the year, more than half (54.1%) reported that this was stable during the past few years. Of the remainder, 38.0% reported working more days and 7.9% reported working fewer.
With regard to the trends in the intensity of work, 24% reported reading the same number of cases per hour, while only 4.2% reported reading fewer, and most, 71.9%, were reading more per hour. When asked how often a participant thought that they were reading faster than they were comfortable with, only around one-fifth (20.5%) reported that reading faster was rare or never occurred. Almost half (44.1%) reported that this occurred sometimes; 29.1% said that this was frequent; and 6.3% of the participants said that this occurred "always." When it came to reports, a similar distribution occurred with 8.1% reporting that they always signed reports faster than they were "comfortable for optimal communication and clarity." When asked if they had too little time to discuss important abnormal results with clinicians, about one-third (32.3%) responded that the lack of time rarely or never happened, while 35.3% said that it occurred sometimes, 27.4% stated it was frequent, and for 5.1% this always happened. In reporting abnormal results, there was a similar distribution, with 4.7% of respondents stating that they always had too little time to report abnormal results.
When asked about reading cases when sleep-deprived, approximately half (48.72%) said this was a rare or never occurrence, while 36.0% reported doing this sometimes, 13.5% did it frequently, and 1.9% did it "always."

On-Call Responsibilities and Perceptions
For the number of duty day/nights of call coverage, 54.1% of respondents reported no change, while in the remainder, those reporting an increase, 36.6%, outnumbered those reporting a decrease, 9.3%. Regarding the perception of on-call duty, only a little more than one-third (35.5%) reported that the difficulty of on call was stable, while of the remainder, those reporting greater difficulty, 59.7%, outnumbered those who thought their on-call duty was easier, 4.9%.
Participants were asked to provide information about what they read on weekend call, defined as Friday evening to Monday morning. The following averages were obtained from 328 respondents: 86 CTs, 44 MRIs, 20 ultrasounds, 93 radiographs, and 13 other (Fig 2), which likely grossly overestimates studies interpreted due to limitations of context in the survey instrument. Important contextual information not obtained as part of the survey included the number of hours on-call, whether calls were fully subspecialized, and whether final or preliminary interpretation was provided, among others.
Of respondents, 62.2% thought that the number of poorly indicated or unindicated studies had increased, while only 4.0% thought that they had decreased, with 33.8% reporting a stable amount.  Number of studies by technique interpreted during a weekend between Friday evening and Monday morning. This is a gross overstatement of individual workload, as these numbers do not account for case number and modality differences between fully subspecialized and partly subspecialized call responsibilities.

Trends in Perceptions of Trainees
Of those who train neuroradiology fellows, most thought that the ability of the trainees was the same as in the past, but of those reporting a change, almost twice as many thought that abilities had declined (Fig 3).
Of respondents who train residents, of those who noticed a change in the past few years, Ͼ4 times as many respondents found a reduction in resident ability than an increase (Fig 4).

Changes in Nonclinical Activities Occurring in Parallel with Increased Clinical Work
Most respondents reported cutting back on Ն1 key nonclinical professional activity (Fig 5 and On-line Table 3) in the past few years. This included key elements of the "academic triad," including teaching and research, as well as mentoring, volunteer work for professional societies, practice building, and personal medical education. Concordant with the earlier results, most of the respondents also reported an impact on the amount of time they have away from work.
Excluding respondents for whom these activities are "Not Applicable," only a small minority of respondents (2 categories combined) never or rarely needed to cut back on the following activities, whereas most (3 categories combined) reported sometimes, frequently, or always cutting back (Fig 5). Regarding cutting back on teaching, of the 60.9% of respondents who teach, 80.0% of those responded that they cut back between "sometimes" and "all the time," with 20.0% who reported never or rarely cutting back. Similar results were obtained regarding the 62.4% of respondents who mentor, 80.1% of whom reported cutting back sometimes or more frequently, with a minority, 19.9%, saying that they never or rarely cut back. Of the 57.9% of respondents who perform and publish research, there was a sharper decrease, with only 12.2% never or rarely cutting back, compared with 87.8% who did. Regarding the 87.8% of respondents who would participate in practice-building activities, 22.4% never or rarely cut back, compared with 77.6% who did. Of the 69.6% of respondents who voluntarily serve in radiology and academic societies, 21.8% of respondents never or rarely cut back on their service compared with 78.2% who did. The 95.6% of respondents who participate in Continuing Medical Education activities were most resistant to cut back, with 38.0% of respondents never or rarely cutting back compared with 62.0% who did. Of the 97.2% of respondents who read journals, 20.7% never or rarely cut back, compared with the 79.3% who did. Most interesting, only 97.2% of respondents reported spending time away from work, and of those, 21.8% never or rarely cut back on nonwork time, compared with 78.2% who did.

Burnout Indices
In answer to 4 commonly considered parameters of burnout or burning out workers, more than half of the respondents said that they had experienced increases in fatigue, difficulty relaxing after work, and diminished enthusiasm   or effectiveness at work. On the fourth question regarding anxiety and depression, slightly less than half reported an increase. When asked if their practice administrators had made changes to help balance clinical and nonclinical work, only about 1 in 7 (14.2%) said yes. More than half the respondents said that they considered retiring earlier than originally planned (51.8%) and 38.9% have considered changing careers. Statistically significant differences were found for several measures divided by experience. Neuroradiologists with Ͻ20 years of experience were more likely to have difficulty relaxing after work (65.1%) versus those with Ն20 years of experience (50.6%, P ϭ .003). Neuroradiologists with Ͻ20 years of experience were more likely to consider retiring earlier than initially intended (57.9%) versus those with Ն20 years of experience (42.4%, P ϭ .002). Neuroradiologists with Ͻ20 years of experience were more likely to consider changing careers (46.0%) versus those with Ն20 years of experience (28.0%, P Ͻ .001).

DISCUSSION
This article reports the results of the first survey of practicing members of the ASNR of their current work, their perceptions of changes in their clinical work, changes in their nonclinical work, their perceptions of their trainees, and 4 commonly used measures of burnout. Radiologists are among the most educated and highly trained professionals in the United States, requiring a minimum of 13 years of education after high school. Neuroradiologists require an additional year (or more) of fellowship training. Staying current requires a strong commitment not only to their work but also to maintaining certification through a combination of Continuing Medical Education and performing service and quality projects. Neuroradiologists thus represent not only a major personal commitment but also a substantial societal investment to create knowledge professionals.
The 21st century has been a time of rapid change in medicine in the United States. Implementation of cost reductions and the shift from volume-to value-based reimbursement have had a strong impact on how radiologists practice. Private practice physicians in many parts of the United States have seen reductions in their autonomy, and others have shifted to employed practice (either voluntary or not); both of these trends have additional impact on the way radiologists practice.
This survey was implemented to begin a periodic process of understanding how neuroradiologists in the United States practice and how they view the quality of their work and its impact. This understanding is important for several reasons.

Sustainability
First, these results suggest crowding out of important nonclinical tasks that academic and private practice radiologists need to do to sustain their careers and their groups/departments. In 2000, Eschelman et al 9 reported an inverse relationship between clinical productivity and academic productivity. If clinical demands crowd out academic work and practice building, then the academic mission will be compromised and the ability of radiologists to build and maintain their practices will be challenged. An important report from a leading academic neuroradiology practice showed that approximately one-fifth of productivity was non-RVU-generating. 10 Second, the sustainability of the radiology enterprise depends on intergenerational work and cooperation. If clinical demands crowd out time for effective teaching, mentoring, and academic project work, the profession itself is at risk. While correlation is not proof of causation, 93.5% (243/260) of our respondents who teach reported a degree of cutting back on teaching, with almost half reporting cutting back frequently or always, 48.1% (125/260). Mentoring activities showed similar degrees of reported reductions. These are professionals acknowledging that they are not doing important tasks for the future of the profession.
The perceptions of neuroradiologists that there has been a decline in the quality of trainees is concerning for the future of the profession and should be investigated further because it could be due to many factors, including the following: generational misunderstandings, burnout in the current teaching cadre leading to overly harsh evaluations, declining educational standards at the medical school level, and/or a decline in the quality of trainees self-selecting into a radiology career.

Potential for Future Workforce Shortage
Although considering early retirement or career change does not necessarily equal intention, it raises the possibility of a shrinking workforce, compounded by recent trends in the general diagnostic radiology National Resident Matching Program.
The historical data from 2004 to 2015, 11-15 years for which information was available at the time this manuscript was written, show a divergence beginning in 2011 in the number of offered training positions from the number of filled positions (Fig 6) in diagnostic radiology residencies. Regarding neuroradiology fellowship positions, there was a milder and delayed divergence between offered positions and filled positions beginning in 2014 (Fig 7), which may be expected, given that fellowship applications occur 4 years after diagnostic residency applications in the third year of residency.
Data from the 2016 National Resident Matching Program, 16,17 however, show an increase in the fill percentage of both diagnostic radiology programs (96.4%) and neuroradiology fellowships (80.5%) compared with the 2015 year (86.3% and 74.5%, respectively), a deviation from the trend of the previous 5 years.
Depending on whether the 2016 data are an anomaly compared with the trailing trend of the past half-decade versus an important inflection point for reversal of the trend, the implication for the availability of future trainees to replace retiring neuroradiologists is great. If the 2016 data are an anomaly, radiology and thus neuroradiology may find itself unable to replace both radiologists who retire as planned when first beginning their careers and any who may retire early or change careers due to burnout. That situation could create a negative cycle, exacerbating the problem. On the other hand, if the 2016 data represent an inflection point, the possibility of a future workforce shortage may be less. More future data points will be needed to evaluate these possibilities.

Burnout
Burnout is a serious threat for this cadre of neuroradiologists. A study of physician happiness comparing specialties reported in 2015 that radiologists were the least happy of 27 types of specialists on a Likert scale. 18 Another report with different methodology also showed radiologists in the lower half of a large group of major medical specialties in reported happiness at work. 19 Burnout is often misunderstood by physicians and administrators. Delay in diagnosis can occur if there is an ignorance of early warning signs. The late indicators include poor work performance, visible anger-yelling, throwing things, verbal abuse, and so forth as well as inappropriate be-havior and medical malpractice. By the time these indicators are manifest, serious personal and professional impairment has already occurred.
Workplace unhappiness extends far beyond the employment setting. It affects relationships, including divorce, and can lead to substance abuse and sleep and eating disorders and neglect of hygiene and health. In a 2015 report, burnout in physicians correlated with the following: a lower rate of volunteerism, worse health, lower exercise, greater likelihood of obesity, less savings, and more likely never being married and being alone. 20 Burnout is also linked to a higher rate of depression; not surprisingly perhaps, the suicide rate among physicians in the United States is higher than that in the general population and is worse for female physicians. 21 Earlier warning signs of burnout include what have been called the "3 D's": disengagement, disinterest, and disconnection. Also, watch for loss of communication, truancy, heightened cynicism, and aggression.
A second problem with accurate diagnosis and treatment of burnout is that burnout is often misunderstood as a character flaw. For example, one may ask why one radiologist cannot work as hard or as long as another? There must be something wrong and so forth. Like posttraumatic stress disorder, burnout can happen to the best of neuroradiologists, despite good character, good training, and so forth. The focus on the individual can also distract leaders and administrators from addressing environmental issues. While some may burn out sooner than others, if a group or department is seeing advanced burnout in one member, that should be a warning to others as well. The same article that noted the level of unhappiness in radiologists compared with other specialties also noted that 10% of radiologists in the United States were severely burned out. 18 In a different article, 49% of radiologists reported at least some burnout. 18,22 Radiology was not the specialty with highest rate of burnout (urology and critical care were the worst), but it was in the top half of specialties examined and reported. 18 Burnout is increasing in the United States. A report in the Mayo Clinic Proceedings from 2011 to 2014 showed that the percentage of physicians reporting at least 1 symptom of burnout increased from 45.5% to 54.4%, while satisfaction with  work-life balance fell from 48.5% to 40.9%. 22 These changes in the medical profession should be considered serious, particularly given that the general population of the United States showed only minimal changes in the same time period. 22 Similarly, another report showed increasing rates of physician burnout from 2013 to 2016, with consistently greater rates in women than men. 19 A recent report from an American College of Radiology Commission focusing on the issue of burnout suggested several remedies for preventing burnout. The highest impact recommendation was to provide adequate staffing. 23

Limitations
This survey has several limitations. It was voluntary, and it is possible that responders are different from the broader neuroradiology community. This is a survey and not an audit of radiologists' work. It is thus open to response and recall biases.
Other authors have noted bias issues in how radiologists and their groups report workload. 2,24 The survey may underestimate changes in workload and difficulty because it does not distinguish between neuroradiologists who work full-time versus neuroradiologists who work part-time or have recently cut back their clinical time.
The number of studies interpreted on-call is likely a gross overexaggeration because it mixes those who take full-subspecialty neuroradiology calls (likely nearly entirely CT and MR imaging) with more general calls (which likely include other modalities beyond CT and MR imaging) and is further confounded by the lack of detail captured in the survey to give those responses context. The survey did not distinguish between preliminary oncall interpretations versus final interpretations or the number of total on-call hours spent interpreting studies, among other confounds.

CONCLUSIONS
This is the first survey in a planned series of surveys of the American Society of Neuroradiology to track changes in radiologists' work, their work environment, and the effects on the quality of their profession and personal lives. Increasing clinical work demands have coincided with destructive effects in the work environment and the ability and desire of radiologists to perform academic or practice-building duties, with a substantial incidence of burnout symptoms in neuroradiologists in the United States. While this survey does not prove causation, the trends and the correlations should be concerning to radiologists, their groups/departments, and their leaders and warrant further monitoring.