- Email: [email protected]
Intensive Care Unit Radiography and the Beginning of the Imaging Value Chain
CASE STUDIES IN CLINICAL PRACTICE MANAGEMENT
Intensive Care Unit Radiography and the Beginning of the Imaging Value Chain Susanna C. Spence, MD, Greta Cardenas, MBA, Bela Patel, MD, Eduardo Matta, MD DESCRIPTION OF THE PROBLEM In a 1,098-bed institution with nine intensive care units, there was a high degree of variability in the response time of our technologists to orders for stat radiography. Our medical intensive care unit (MICU) colleagues came to us dissatisfied with our response times, stating that the information on those images was needed for urgent patient care decisions and that we were not currently meeting their expectations. The radiographic report was not their primary concern, as they felt comfortable with their preliminary interpretations of intensive care unit (ICU) radiographs for life-threatening conditions. However, obtaining the image itself as fast as possible was critically important to them. At the start of the study, the median time it took a technologist to reach the bedside to begin a stat examination in the MICU was 46 min 46 seconds after the order was placed. Our MICU colleagues told us that this was unacceptable. WHAT WAS DONE Baseline data were obtained from the radiology information system and were analyzed at the end of November 2013 for the previous 12 months (December 2012 to November 2013). After some discussion, our MICU colleagues told us that the metric they would like us to measure
was the time it took a technologist to reach a patient’s bedside, as they realized that many patient-specific factors might delay the completion of the image after the technologist had arrived. We therefore defined this metric (order to start) as our radiographic turnaround time (TAT) for the purposes of the project. After being told that our baseline TAT measurement of 46 min 46 seconds for stat radiography was not acceptable, we asked our MICU colleagues what length of time would meet their needs. The attending physicians in the MICU believed that a stat radiographic examination was a rare event, but a critical one, and should be treated as such. They told us that they needed a technologist to arrive at bedside for all stat radiographic examinations within 15 min, without exceptions. Establishing that as a demanding target given the current median, we began our project. To work within a reasonable time frame and on a reasonable scale, the scope of this project was initially limited to the MICU. We also limited the project to portable chest and abdominal radiography; stat foot radiography would not be held to the 15-min TAT. We started by simply walking the process, establishing the steps and flow of the information from the time an order was placed until the image was sent to the PACS. Following the technologist around
ª 2015 American College of Radiology 1546-1440/15/$36.00 n http://dx.doi.org/10.1016/j.jacr.2015.06.028
while morning radiographs were taken, watching the clinicians place an order, and speaking with everyone from MICU nurses to radiology clerks shed a lot of light on the workflow [1]. Technologist surveys, feedback from clinicians, and data analysis from the radiology information system eventually led us to identify three major problems we could address: n
n
n
a traditional culture of ordering that resulted in marked overuse of the term stat, outdated equipment that was often physically located far from where it was needed, and lack of timely communication of the stat order to the technologist.
The MICU attending physicians felt that a true stat radiographic examination was a rare event, perhaps occurring three or four times a day. From a practical standpoint, however, the number of stat examinations ordered was often much greater. The term stat was creeping into orders placed for tomorrow’s early-morning chest radiographic study or “follow-up pneumothorax” radiography for 6 hours from now. Looking at our 12-month baseline data, we found clinicians who had placed one stat order during the whole year and others who had placed 103 stat orders during a 1-month rotation. This variability
1
came from multiple sources, including uncertainty about when a routine might be performed and a fundamental belief that “it’s an ICU case, therefore it’s stat.” This gave rise to our first intervention: order patterns.
Intervention 1: Ordering The radiologists on the team did not try to dictate what a stat radiographic examination was to the MICU physicians. We defined a true stat versus a false stat examination quite simply: a true stat study had to be ordered for, and needed, right now. Any radiographic examination that was ordered for some time in the future and marked “stat” was considered false. Typical baseline data are shown in Figure 1 for October 2013. At the beginning of December 2013, the lead author attempted to change order patterns singlehandedly, by directly informing the residents and fellows on the
MICU service that the processes had changed and that the term stat should be used judiciously, and only for examinations absolutely needed immediately. The complete lack of effect of this intervention, and in fact worsening of the problem, was shown by the data analysis at the end of that month. It became clear that alterations in MICU culture and workflow could not be effected by what was perceived as a purely external source [2]. Therefore, starting in January 2014, the lead author would drop by early-morning rounds near the first day of the rotation. The MICU attending physician on service would personally pull all the residents and fellows aside, have them listen to the radiologist’s presentation on the need to limit the number of stat orders to critical studies needed now, and then verbally reinforce his or her support for the new workflow. The change in
order patterns was immediate and long-standing. This workflow has been holding well in the MICU culture, and the introduction is in the hands of the MICU charge nurse during orientation.
Intervention 2: Equipment During our analysis, we found that the portable equipment being used was often shared between different towers of the building. The battery life on the machine shared by the MICU tower was dismal (w12-14 images before the battery died and the unit had to be recharged), it took 6.3 min to boot up, and a tethered cord required a full wipe-down after each bed, slowing the technologist down. The TAT data and documentation of workflow for this unit and others were used to help show that the equipment itself resulted in delays. Additional equipment was bought for the hospital as a whole,
Fig 1. Average number of stat radiographic examinations per day, divided over the month. “False” stat orders were defined as stat examinations ordered for some time in the future, such as chest radiography to follow up pneumothorax in 6 hours. True stat examinations were defined as those ordered to be performed immediately.
2
Journal of the American College of Radiology Volume - n Number - n - 2015
with one unit now dedicated to the tower containing the MICU.
Intervention 3: Communication We also surveyed our technologists on every case for 20 days, asking if they made it to the bedside within 15 min, and if not, why not. The survey showed a lack of communication of orders as the most common cause of delay. When we looked into how orders were received by the technologist, we discovered that the person ordering a study had to either page the technologist personally or had to request that the nurse, front desk staff member, or medical student page the technologist. On the basis of pager audits, this occurred successfully only about 38% of the time, with a median and mean delay of 5 and 9.5 min, respectively. The rest of the time, no one paged the technologist at all, and the study was discovered on the work list the next time the technologist had a moment to check it. A lot of our 15 min (and indeed much more than 15 min in many cases) was already gone. Borrowing an idea from the pediatric ICU in the same hospital, we instituted a portable phone to obviate the need for a callback from the technologist
and to give the clinicians more ways to access the technologist. Unfortunately, this was not effective, as the portable phone simply replaced the pager on a few cases, and many examinations still received neither page nor call. Median TATs did improve by approximately 2 min during the pilot phase, but this was not nearly effective enough as a solution. Still, when we analyzed the pager data, it was clear that the technologist was faster at getting to the bedside when paged (Fig. 2). Therefore, because we knew that education is a short-lived intervention, we decided to take the human element out altogether. We went to our IT department and asked them to help us automate the paging system. A stat order placed in the MICU now automatically goes to the technologist pager: for example, “MICU Bed 04 Stat.”
OUTCOMES The numeric results of the three adopted interventions are shown serially in Figure 3. Each of these effectively represents a plan-do-study-act cycle [3]. The abandoned pilot phase with the portable phone is not shown. Median TAT decreased from 46 min 46 seconds at baseline, to 10 min
8 seconds after all three interventions, and 9 min 18 seconds averaged over the postintervention months. Unfortunately, a median of 15 min was not what our ICU colleagues had asked for. They wanted all radiographic examinations that were stat performed within 15 min; therefore, the metric that is tracked and placed monthly on the radiology technologist’s wall is very simple: percentage of radiographs meeting the 15-min TAT. This is a strict requirement, and we do not always meet it, although some of our failures are explained by the patient or physician not being ready when the technologist arrives, so that the technologist is therefore not able to start the examination. There is always more work to be done, but the difference in the MICU is palpable to our physicians. At the start of the study, our first inclination was to solve the TAT problem immediately, simply by hiring more technologists for all four shifts. However, after we actually spent some time finding out what was happening on the ground, and mapping the process step by step, we found that the problems we encountered had nothing to do with the number of technologists
Fig 2. Percentage of total cases each month in which the technologist reached the bedside within the 15-min window, expressed as all cases versus cases in which the technologist was paged.
Journal of the American College of Radiology Spence et al n Case Studies in Clinical Practice Management
3
Fig 3. Median turnaround times and distribution of cases for the three major phases of the study, each effectively representing a plan-do-study-act cycle. For graphical representation, 2 of the 12 preintervention months, and 3 of the 6 postintervention months are shown.
available on a given day. Not a single additional technologist was hired to accomplish an approximately 80% reduction in TAT. Equipment was bought, and although it had been generally known that a purchase would need to occur at some point, the Lean tools and the fact that we were not meeting clinicians’ expectations allowed us to make a stronger case for the purchase. Decreasing the number of stat orders in the first place to allow appropriate prioritization, and then finally taking the human factor out of communications, ended up being key factors. This kind of initiative is eminently
generalizable, with the system now running in three of our nine ICUs, with more on the way. For our ordering clinicians, the clock did not start once an examination hit the radiologist’s work list, it started the moment they placed the order. Ignoring the need for speed and accuracy in this part of the process meant a lot of dissatisfaction from our referrers, even though their professional relationship with the chest radiologists was both positive and long-standing [4]. Although the quality of the final report was simply assumed and implied, the face of the radiology department, seen every day,
was the technologist, whether immediately or after a lengthy wait. We look forward to continuing to expand this initiative to our remaining units and across other hospitals in the system.
REFERENCES 1. Kruskal JB, Reedy A, Pascal L, Rosen MP, Boiselle PM. Quality initiatives: Lean approach to improving performance and efficiency in a radiology department. Radiographics 2012;32:573-87. 2. Kotter JP. Leading change. Boston, Massachusetts: Harvard Business Review Press; 2012. 3. Tague NR. The quality toolbox. 2nd ed. Milwaukee, Wisconsin: ASQ Quality Press; 2005. 4. Ellenbogen P. Imaging 3.0: what is it? J Am Coll Radiol 2013;10:229.
Susanna C. Spence, MD, and Eduardo Matta, MD, are from the Department of Diagnostic and Interventional Imaging, University of Texas Medical School at Houston, Houston, Texas. Greta Cardenas, MBA, is from the Department of Diagnostic and Interventional Imaging, Memorial Hermann e Texas Medical Center, Houston, Texas. Bela Patel, MD, is from the Department of Internal Medicine, Division of Critical Care, University of Texas Medical School at Houston, Houston, Texas. The authors have no conflicts of interest related to the material discussed in this article. Susanna C. Spence, MD: Department of Diagnostic and Interventional Imaging, University of Texas Medical School at Houston, 6431 Fannin Street, MSB 2.130B, Houston, TX 77030; e-mail: [email protected].
4
Journal of the American College of Radiology Volume - n Number - n - 2015
Intensive Care Unit Radiography and the Beginning of the Imaging Value Chain Susanna C. Spence, MD, Greta Cardenas, MBA, Bela Patel, MD, Eduardo Matta, MD DESCRIPTION OF THE PROBLEM In a 1,098-bed institution with nine intensive care units, there was a high degree of variability in the response time of our technologists to orders for stat radiography. Our medical intensive care unit (MICU) colleagues came to us dissatisfied with our response times, stating that the information on those images was needed for urgent patient care decisions and that we were not currently meeting their expectations. The radiographic report was not their primary concern, as they felt comfortable with their preliminary interpretations of intensive care unit (ICU) radiographs for life-threatening conditions. However, obtaining the image itself as fast as possible was critically important to them. At the start of the study, the median time it took a technologist to reach the bedside to begin a stat examination in the MICU was 46 min 46 seconds after the order was placed. Our MICU colleagues told us that this was unacceptable. WHAT WAS DONE Baseline data were obtained from the radiology information system and were analyzed at the end of November 2013 for the previous 12 months (December 2012 to November 2013). After some discussion, our MICU colleagues told us that the metric they would like us to measure
was the time it took a technologist to reach a patient’s bedside, as they realized that many patient-specific factors might delay the completion of the image after the technologist had arrived. We therefore defined this metric (order to start) as our radiographic turnaround time (TAT) for the purposes of the project. After being told that our baseline TAT measurement of 46 min 46 seconds for stat radiography was not acceptable, we asked our MICU colleagues what length of time would meet their needs. The attending physicians in the MICU believed that a stat radiographic examination was a rare event, but a critical one, and should be treated as such. They told us that they needed a technologist to arrive at bedside for all stat radiographic examinations within 15 min, without exceptions. Establishing that as a demanding target given the current median, we began our project. To work within a reasonable time frame and on a reasonable scale, the scope of this project was initially limited to the MICU. We also limited the project to portable chest and abdominal radiography; stat foot radiography would not be held to the 15-min TAT. We started by simply walking the process, establishing the steps and flow of the information from the time an order was placed until the image was sent to the PACS. Following the technologist around
ª 2015 American College of Radiology 1546-1440/15/$36.00 n http://dx.doi.org/10.1016/j.jacr.2015.06.028
while morning radiographs were taken, watching the clinicians place an order, and speaking with everyone from MICU nurses to radiology clerks shed a lot of light on the workflow [1]. Technologist surveys, feedback from clinicians, and data analysis from the radiology information system eventually led us to identify three major problems we could address: n
n
n
a traditional culture of ordering that resulted in marked overuse of the term stat, outdated equipment that was often physically located far from where it was needed, and lack of timely communication of the stat order to the technologist.
The MICU attending physicians felt that a true stat radiographic examination was a rare event, perhaps occurring three or four times a day. From a practical standpoint, however, the number of stat examinations ordered was often much greater. The term stat was creeping into orders placed for tomorrow’s early-morning chest radiographic study or “follow-up pneumothorax” radiography for 6 hours from now. Looking at our 12-month baseline data, we found clinicians who had placed one stat order during the whole year and others who had placed 103 stat orders during a 1-month rotation. This variability
1
came from multiple sources, including uncertainty about when a routine might be performed and a fundamental belief that “it’s an ICU case, therefore it’s stat.” This gave rise to our first intervention: order patterns.
Intervention 1: Ordering The radiologists on the team did not try to dictate what a stat radiographic examination was to the MICU physicians. We defined a true stat versus a false stat examination quite simply: a true stat study had to be ordered for, and needed, right now. Any radiographic examination that was ordered for some time in the future and marked “stat” was considered false. Typical baseline data are shown in Figure 1 for October 2013. At the beginning of December 2013, the lead author attempted to change order patterns singlehandedly, by directly informing the residents and fellows on the
MICU service that the processes had changed and that the term stat should be used judiciously, and only for examinations absolutely needed immediately. The complete lack of effect of this intervention, and in fact worsening of the problem, was shown by the data analysis at the end of that month. It became clear that alterations in MICU culture and workflow could not be effected by what was perceived as a purely external source [2]. Therefore, starting in January 2014, the lead author would drop by early-morning rounds near the first day of the rotation. The MICU attending physician on service would personally pull all the residents and fellows aside, have them listen to the radiologist’s presentation on the need to limit the number of stat orders to critical studies needed now, and then verbally reinforce his or her support for the new workflow. The change in
order patterns was immediate and long-standing. This workflow has been holding well in the MICU culture, and the introduction is in the hands of the MICU charge nurse during orientation.
Intervention 2: Equipment During our analysis, we found that the portable equipment being used was often shared between different towers of the building. The battery life on the machine shared by the MICU tower was dismal (w12-14 images before the battery died and the unit had to be recharged), it took 6.3 min to boot up, and a tethered cord required a full wipe-down after each bed, slowing the technologist down. The TAT data and documentation of workflow for this unit and others were used to help show that the equipment itself resulted in delays. Additional equipment was bought for the hospital as a whole,
Fig 1. Average number of stat radiographic examinations per day, divided over the month. “False” stat orders were defined as stat examinations ordered for some time in the future, such as chest radiography to follow up pneumothorax in 6 hours. True stat examinations were defined as those ordered to be performed immediately.
2
Journal of the American College of Radiology Volume - n Number - n - 2015
with one unit now dedicated to the tower containing the MICU.
Intervention 3: Communication We also surveyed our technologists on every case for 20 days, asking if they made it to the bedside within 15 min, and if not, why not. The survey showed a lack of communication of orders as the most common cause of delay. When we looked into how orders were received by the technologist, we discovered that the person ordering a study had to either page the technologist personally or had to request that the nurse, front desk staff member, or medical student page the technologist. On the basis of pager audits, this occurred successfully only about 38% of the time, with a median and mean delay of 5 and 9.5 min, respectively. The rest of the time, no one paged the technologist at all, and the study was discovered on the work list the next time the technologist had a moment to check it. A lot of our 15 min (and indeed much more than 15 min in many cases) was already gone. Borrowing an idea from the pediatric ICU in the same hospital, we instituted a portable phone to obviate the need for a callback from the technologist
and to give the clinicians more ways to access the technologist. Unfortunately, this was not effective, as the portable phone simply replaced the pager on a few cases, and many examinations still received neither page nor call. Median TATs did improve by approximately 2 min during the pilot phase, but this was not nearly effective enough as a solution. Still, when we analyzed the pager data, it was clear that the technologist was faster at getting to the bedside when paged (Fig. 2). Therefore, because we knew that education is a short-lived intervention, we decided to take the human element out altogether. We went to our IT department and asked them to help us automate the paging system. A stat order placed in the MICU now automatically goes to the technologist pager: for example, “MICU Bed 04 Stat.”
OUTCOMES The numeric results of the three adopted interventions are shown serially in Figure 3. Each of these effectively represents a plan-do-study-act cycle [3]. The abandoned pilot phase with the portable phone is not shown. Median TAT decreased from 46 min 46 seconds at baseline, to 10 min
8 seconds after all three interventions, and 9 min 18 seconds averaged over the postintervention months. Unfortunately, a median of 15 min was not what our ICU colleagues had asked for. They wanted all radiographic examinations that were stat performed within 15 min; therefore, the metric that is tracked and placed monthly on the radiology technologist’s wall is very simple: percentage of radiographs meeting the 15-min TAT. This is a strict requirement, and we do not always meet it, although some of our failures are explained by the patient or physician not being ready when the technologist arrives, so that the technologist is therefore not able to start the examination. There is always more work to be done, but the difference in the MICU is palpable to our physicians. At the start of the study, our first inclination was to solve the TAT problem immediately, simply by hiring more technologists for all four shifts. However, after we actually spent some time finding out what was happening on the ground, and mapping the process step by step, we found that the problems we encountered had nothing to do with the number of technologists
Fig 2. Percentage of total cases each month in which the technologist reached the bedside within the 15-min window, expressed as all cases versus cases in which the technologist was paged.
Journal of the American College of Radiology Spence et al n Case Studies in Clinical Practice Management
3
Fig 3. Median turnaround times and distribution of cases for the three major phases of the study, each effectively representing a plan-do-study-act cycle. For graphical representation, 2 of the 12 preintervention months, and 3 of the 6 postintervention months are shown.
available on a given day. Not a single additional technologist was hired to accomplish an approximately 80% reduction in TAT. Equipment was bought, and although it had been generally known that a purchase would need to occur at some point, the Lean tools and the fact that we were not meeting clinicians’ expectations allowed us to make a stronger case for the purchase. Decreasing the number of stat orders in the first place to allow appropriate prioritization, and then finally taking the human factor out of communications, ended up being key factors. This kind of initiative is eminently
generalizable, with the system now running in three of our nine ICUs, with more on the way. For our ordering clinicians, the clock did not start once an examination hit the radiologist’s work list, it started the moment they placed the order. Ignoring the need for speed and accuracy in this part of the process meant a lot of dissatisfaction from our referrers, even though their professional relationship with the chest radiologists was both positive and long-standing [4]. Although the quality of the final report was simply assumed and implied, the face of the radiology department, seen every day,
was the technologist, whether immediately or after a lengthy wait. We look forward to continuing to expand this initiative to our remaining units and across other hospitals in the system.
REFERENCES 1. Kruskal JB, Reedy A, Pascal L, Rosen MP, Boiselle PM. Quality initiatives: Lean approach to improving performance and efficiency in a radiology department. Radiographics 2012;32:573-87. 2. Kotter JP. Leading change. Boston, Massachusetts: Harvard Business Review Press; 2012. 3. Tague NR. The quality toolbox. 2nd ed. Milwaukee, Wisconsin: ASQ Quality Press; 2005. 4. Ellenbogen P. Imaging 3.0: what is it? J Am Coll Radiol 2013;10:229.
Susanna C. Spence, MD, and Eduardo Matta, MD, are from the Department of Diagnostic and Interventional Imaging, University of Texas Medical School at Houston, Houston, Texas. Greta Cardenas, MBA, is from the Department of Diagnostic and Interventional Imaging, Memorial Hermann e Texas Medical Center, Houston, Texas. Bela Patel, MD, is from the Department of Internal Medicine, Division of Critical Care, University of Texas Medical School at Houston, Houston, Texas. The authors have no conflicts of interest related to the material discussed in this article. Susanna C. Spence, MD: Department of Diagnostic and Interventional Imaging, University of Texas Medical School at Houston, 6431 Fannin Street, MSB 2.130B, Houston, TX 77030; e-mail: [email protected].
4
Journal of the American College of Radiology Volume - n Number - n - 2015