- Email: [email protected]
An Analytic Decision Support Tool for Resident Allocation
ORIGINAL REPORTS
An Analytic Decision Support Tool for Resident Allocation Is¸ılay Talay-Deg ˘ irmenci, PhD,* Casey J. Holmes, MD,† Paul C. Kuo, MD,† and Otis B. Jennings, PhD‡ *Department of Business Administration, College of Business, Antalya International University, Antalya, Turkey; † Department of Surgery, Loyola University Medical Center, Maywood, Illinois; and ‡Duke University, The Fuqua School of Business, Durham, North Carolina BACKGROUND: Moving residents through an academic residency program is complicated by a number of factors. Across all residency programs the percentage of residents that leave for any reason is between 3.4% and 3.8%.1 There are a number of residents that participate in research. To avoid discrepancies in the number of residents at the all levels, programs must either limit the number of residents that go into the lab, the number that return to clinical duties, or the number of interns to hire. Traditionally this process consists of random selection and trial and error with names on magnetic strips moved around a board. With the matrix that we have developed this process is optimized and aided by a Microsoft Excel macro (Microsoft Corp, Redmond, Washington). METHODS: We suggest that a residency program would have
the same number of residents at each residency stage of clinical practice, as well as a steady number of residents at each research stage. The program consists of 2 phases, in the first phase, an Excel sheet called the “Brain Sheet,” there are simple formulas that we have prepared to determine the number of interns to recruit, residents in the research phase, and residents that advance to the next stage of training. The second phase of the program, the macro, then takes the list of current resident names along with the residency level they are in, and according to the formulas allocates them to the relevant stages for future years, creating a resident matrix. RESULTS: Our macro for resident allocation would maximize the time of residency program administrators by simplifying the movement of residents through the program. It would also provide a tool for planning the number of new interns to recruit and program expansion. CONCLUSIONS: The application of our macro illustrates that analytical techniques can be used to minimize the time
Correspondence: Address inquiries to Paul C. Kuo, MD, MBA, LUMC, Department of Surgery, EMS Building, RM 3244, 2160 South First Ave., Maywood, IL 60153; fax: 708-327-2852; e-mail: [email protected]
spent and avoid the trial and error while planning resident movement in a program. (J Surg 70:31-35. © 2012 Association of Program Directors in Surgery. Published by Elsevier Inc. All rights reserved.) KEY WORDS: operations research, surgery residency, macro,
schedule COMPETENCIES: Patient Care, Practice-Based Learning and Improvement, Systems-Based Practice
INTRODUCTION Academic residency programs fulfill the important mission of training residents as clinical practitioners with specialized expertise. There are a large portion of residency programs that offer a dedicated time for research and others require research experience for completion of program.1 The duration of residency programs in surgery is at least 5 years; those that require research are 7 years. Residents that participate in research traditionally work in the clinical setting for the first 2 years and then leave for their research elective.2,3 These residents spend 2 or more years doing research or earning a PhD, in some cases. Finally, these residents then spend 3 years in clinical training before graduating from the program. Differences in this path may occur, but we suggest our model can be modified according to the specifications of a residency program. Across all residency programs the percentage of residents that leave the program is between 3.4% and 3.8%.4 The number of those who will take a break from clinical duties for a research elective compared with the number of residents that have completed their time in the lab and will return might also vary. The combination of all these factors can lead to a discrepancy in the number of residents at each level of training.5,6 The objective of this report is to provide analytically guided recommendations to the administrators of a surgery residency program for determining the size of the postgraduate year (PGY)-1 class and the number of residents that will continue to the PGY-3 each year. When discrepancies from the ideal number of trainees at each level occur because of attrition and other
Journal of Surgical Education • © 2012 Association of Program Directors in Surgery 1931-7204/$30.00 Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jsurg.2012.07.003
31
factors do occur, this model will allow a residency program to reach the ideal flow in as few years as possible, and at the same time abide by the necessary minimum and maximum number of chief residents (CR) and the minimum number of interns (I). We provide, to our knowledge, a first step to transform this process into an analytically guided method where the current resident names and the number of the residents at each level are entered into an Excel workbook along with the ideal PGY and research year (RY) levels, minimum and maximum CR, minimum I, and the Excel macro yields a resident allocation matrix with names and projections for future years.
METHODS The URL for the Resident Matrix Calculator is https://dl. dropbox.com/u/45447844/Resident%20Matrix%20Calculator. xlsm. Alternatively, the senior author can be contacted. With a significant number of residents taking advantage of research electives or leaving their training program, the spaces left at each PGY level must be filled. Administrators of a program may have to choose the minimum and maximum number of residents required at each PGY level. For example, the last PGY level, the CR, might be required to have a number of residents between predetermined minimum and maximum numbers; and the first PGY level, I, may have a minimum number of spaces required to staff the program. Research is assumed not to be mandatory for our model; therefore it is expected that the number of residents choosing to do research during residency may change from year to year. Financial considerations, such as research budget restraints and the availability of grants also plays a role in the decision of the number or residents allowed to enter the lab and can change annually. Technical considerations, such as equipment and lab space availability are also among the determining factors for the number of spaces open in the research labs. There are 2 decisions the administrators of a residency program make annually: how many interns should be recruited, and how many PGY-2 residents (JARs) must proceed directly to the next clinical training stage, PGY-3 (SAR1). The administrators of the residency programs fortunately have a certain level of control on part of the system dynamics. First of all, they determine the number of I to hire each year. Second, by also considering the financial and regulatory constraints, they can also determine the number of residents that continue clinical practice without participating in research, ie, these residents will proceed to PGY-3 directly from PGY-2 and will not be in the research year (RY) status. Residents at 1 PGY level are expected to advance to the next PGY level the following year; however, the above factors create the desire to avoid large fluctuations among the number of residents at every PGY level. The objective of the formulas is to reach the desired balance for the allocation of the residents in the shortest time possible starting from the current situation. The formulas are created using combinations of the 4 operations and min() and max() built-in functions of Microsoft Excel, and they give the number of interns that need to be recruited and the number of JARs that 32
may proceed to SAR1 each year. In the Excel workbook containing the macro, our formulas are placed in the Excel worksheet called the “Brain Sheet.” In the “Brain Sheet,” the first row of the table is spared to enter the numbers for the current year’s distribution of residents. The user enters the current year and how many residents are at each stage of the residency program. The formulas are in the second row. Thus, after entering the first row, the (already entered) formulas automatically calculate the following year’s distribution and hence shows the number of I to recruit and JARs to proceed to SAR1 the next year, and all the user must do is “pull down” the formulas on the second line and they give the future years’ allocation of residents. If enough rows are created by “pulling down” the formulas, the user will eventually see the desired balance in resident allocation. Our formulas are created so that this will be achieved in the least number of years possible. To increase the versatility we added 3 constraints that the user may enforce; the maximum number of CR, minimum number of CR, and the minimum number of I. If there are no such constraints, then these constraints can be relaxed by entering 0 for the minimum number required and a very large number which will not be reached in reality for the maximum number (eg, 10,000). The program consists of 2 phases; in the first phase, the Excel worksheet called the “Brain Sheet” includes formulas that we prepared to calculate/do the following for future years (Fig. 1): 1. the number of I (PGY-1) to recruit; 2. the number of JARs (PGY-2) that will bypass research and progress to the SAR1 (PGY-3) level; 3. the number of JARs that will enter the research year 1tract; 4. The program will also distribute the correct number of residents to the research year 2, SAR1 (PGY-3), PGY-4 (SAR2), and CR (PGY-5) levels. For residents who would like to advance in their research expertise we also added a research year x (RFx) state where it is possible for residents to enter and stay as long as desired, for example if they want to obtain a PhD. The second phase of the program consists of an Microsoft Excel macro created via Visual Basic integrated design environment, which then takes the list of current resident names along with the residency state they are in, and according to the formulas listed in the “Brain Sheet,” allocates the residents to the relevant stages for future years, creating a resident matrix that shows the annual assignment planning, starting from the current year (Fig. 2). Excel Macro To make full use of the program the user enters the names of the current residents into the “Entry Sheet” and the user enters minimum CR, maximum CR, minimum I, desired value (DV), and steady research fellow (RF) on the “Brain Sheet.” The user then enters the current year and the number of residents at each stage of the current year into the first row of the table in the “Brain Sheet” and then the already entered formulas at the second row automatically give recommended numbers for next
Journal of Surgical Education • Volume 70/Number 1 • January/February 2013
Desired Value 7
Years 2011 2012 2013 2014 2015 2016 2017
Start 7 5 6 7 7 7 7
Maximum CR 8
Interns End_JAR 7 5 6 7 7 7 7
Loss
Minimum Interns 5
Min CR 7
Steady RF 3
For People Doing PhDs JARs RF1 RF2 RFx SAR1 SAR2 CR Start End_SAR1 Loss End_RF1 End_RF2 End_RFx Start End_RF2 Loss End_RFx End_SAR1 Start End_SAR1 Loss End_RFx Start End_SAR1 End_RFx Loss Start End_SAR2 Loss Start End_CR Loss Start End Loss 7 2 5 5 5 6 6 8 8 7 7 7 7 7 3 4 5 5 5 5 0 0 8 8 8 8 7 7 5 2 3 4 4 5 5 0 0 8 8 8 8 8 8 6 3 3 3 3 4 4 0 0 7 7 8 8 8 8 7 4 3 3 3 3 3 0 0 7 7 7 7 8 8 7 4 3 3 3 3 3 0 0 7 7 7 7 7 7 7 4 3 3 3 3 3 0 0 7 7 7 7 7 7
FIGURE 1. An example of the “Brain Sheet.” in which a sample starting resident allocation and parameter set is entered and manipulated. CR, chief resident or PGY-5; Intern, PGY-1; JAR, PGY-2; PGY, postgraduate year; RF1, research fellow first year; RF2, research fellow second year; SAR1, PGY-3; SAR2, PGY-4.
year’s PGY and RY levels. “Pulling down” the formulas at the second row until resident allocation has the desired balance will give the recommended projection to the administrators.
INT
JAR
RF 1
RF 2
SAR1
SAR2
CR
Y ear 2011-2012 .2011.1 .2011.2 .2011.3 .2011.4 .2011.5 .2011.6 .2011.7 .2010.1 .2010.2 .2010.3 .2010.4 .2010.5 .2010.6 .2010.7 .2009.1 .2009.2 .2009.3 .2009.4 .2009.5 .2008.1 .2008.2 .2008.3 .2008.4 .2008.5 .2008.6 .2007.1 .2007.2 .2007.3 .2007.4 .2007.5 .2007.6 .2007.7 .2007.8 .2006.1 .2006.2 .2006.3 .2006.4 .2006.5 .2006.6 .2006.7 .2005.1 .2005.2 .2005.3 .2005.4 .2005.5 .2005.6 .2005.7
INT
Year 2012-2013 2012. 1 2012. 2 2012. 3 2012. 4 2012. 5
JAR
INT
.2011.1 .2011.2 .2011.3 .2011.4 .2011.5 .2011.6 .2011.7 RF 1 .2010.1 .2010.2 .2010.3 .2010.4 .2010.5 RF 2 .2009.1 .2009.2 .2009.3 .2009.4 .2009.5
JAR
SAR 1 .2010.6 .2010.7 .2008.1 .2008.2 .2008.3 .2008.4 .2008.5 .2008.6 SAR2 .2007.1 .2007.2 .2007.3 .2007.4 .2007.5 .2007.6 .2007.7 .2007.8 CR .2006.1 .2006.2 .2006.3 .2006.4 .2006.5 .2006.6 .2006.7
Y ear 2013-2014 2013. 1 2013. 2 2013. 3 2013. 4 2013. 5 2013. 6 2012. 2012. 2012. 2012. 2012.
1 2 3 4 5
INT
JAR
After enabling the macros, when the “Front Sheet” is clicked, the macro automatically creates the resident annual assignment planning matrix using the names from the “Entry Sheet.” The Year 2014-2015 2014. 1 2014. 2 2014. 3 2014. 4 2014. 5 2014. 6 2014. 7 2013. 1 2013. 2 2013. 3 2013. 4 2013. 5 2013. 6 2012. 1 2012. 2 2012. 3
Year 2015-2016 2015. 1 2015. 2 2015. 3 2015. 4 2015. 5 2015. 6 2015. 7 JAR 2014. 1 2014. 2 2014. 3 2014. 4 2014. 5 2014. 6 2014. 7 RF 1 2013. 1 2013. 2 2013. 3
INT
Year 2016-2017 2016. 1 2016. 2 2016. 3 2016. 4 2016. 5 2016. 6 2016. 7 JAR 2015. 1 2015. 2 2015. 3 2015. 4 2015. 5 2015. 6 2015. 7 RF 1 2014. 1 2014. 2 2014. 3
INT
Year 2017-2018 2017. 1 2017. 2 2017. 3 2017. 4 2017. 5 2017. 6 2017. 7 JAR 2016. 1 2016. 2 2016. 3 2016. 4 2016. 5 2016. 6 2016. 7 RF 1 2015. 1 2015. 2 2015. 3
INT
RF 1 .2011.1 . 2011. 2 . 2011. 3 . 2011. 4
RF 1
RF 2 .2010.1 . 2010. 2 . 2010. 3 . 2010. 4 . 2010. 5
RF 2 .2011.1 . 2011. 2 . 2011. 3 . 2011. 4
RF 2
SAR 1 .2011.5 . 2011. 6 . 2011. 7 . 2009. 1 . 2009. 2 . 2009. 3 . 2009. 4 . 2009. 5 SAR2 .2010.6 . 2010. 7 . 2008. 1 . 2008. 2 . 2008. 3 . 2008. 4 . 2008. 5 .2008.6 CR .2007.1 .2007. 2 .2007. 3 .2007. 4 .2007. 5 .2007. 6 .2007. 7 .2007.8
SAR 1 2012. 4 2012. 5 . 2010. 1 . 2010. 2 . 2010. 3 . 2010. 4 . 2010. 5
SAR 1 2013. 4 2013. 5 2013. 6 . 2011.1 . 2011.2 . 2011.3 . 2011.4
SAR 1 2014. 2014. 2014. 2014. 2012. 2012. 2012.
4 5 6 7 1 2 3
SAR 1 2015. 2015. 2015. 2015. 2013. 2013. 2013.
4 5 6 7 1 2 3
SAR2 .2011.5 . 2011. 6 . 2011. 7 . 2009. 1 . 2009. 2 . 2009. 3 . 2009. 4 .2009.5 CR .2010.6 . 2010. 7 . 2008. 1 . 2008. 2 . 2008. 3 . 2008. 4 . 2008. 5 .2008.6
SAR2 2012. 4 2012. 5 . 2010.1 . 2010.2 . 2010.3 . 2010.4 . 2010.5
SAR2 2013. 4 2013. 5 2013. 6 .2011.1 .2011.2 .2011.3 .2011.4
SAR2 2014. 2014. 2014. 2014. 2012. 2012. 2012.
4 5 6 7 1 2 3
CR
CR
CR
2012. 1 2012. 2 2012. 3
.2011.5 .2011.6 .2011.7 .2009.1 .2009.2 .2009.3 .2009.4 .2009.5
RF 2
2013. 1 2013. 2 2013. 3
2012. 4 2012. 5 .2010. 1 .2010. 2 .2010. 3 .2010. 4 .2010. 5
RF 2
2014. 1 2014. 2 2014. 3
2013. 4 2013. 5 2013. 6 .2011. 1 .2011. 2 .2011. 3 .2011. 4
FIGURE 2. An example of the “Front Sheet” or the matrix of residents and how they are predicted to progress through the training program based on the formulas contained in the “Brain sheet” and names entered on the “Entry Sheet.” CR, chief resident or PGY-5; Intern, PGY-1; JAR, PGY-2; PGY, postgraduate year; RF1, research fellow first year; RF2, research fellow second year; SAR1, PGY-3; SAR2, PGY-4. Journal of Surgical Education • Volume 70/Number 1 • January/February 2013
33
macro created for demonstrational purposes of this project by default starts placing the JARs to first year RF from the top of the list until the desired amount dictated by the formula is reached, then places the rest of the JARs to SAR1 for next year. When resident names are entered, it could be more suitable to list those who are more likely to choose research electives at the top of their respective list corresponding to their residency stage.
DISCUSSION Moving residents through an academic residency program is a complicated process and takes a considerable amount of time and resources. All residency programs deal with attrition and many deal with residents taking time from clinic duties to participate in research. These issues are further complicated by expansion of the program, and varying numbers of residents that leave for the lab versus the number returning. The combination of these factors leads to a discrepancy in the number of residents at each level of training. To avoid discrepancies in the number of CR, programs either limit the number of residents that go into the lab or the number that directly continues to clinical duties. The program must also consider the size of the PGY-1 class and the number of residents that have left the program. Traditionally, the process of selecting residents to move into and out of the lab consists of random trial and error. With the macro that we have developed this process is transformed from error-prone guess work to an analytically guided decision process. The macro actually was designed for a program with mandatory research years of a prescribed duration. Situations arose in which residents needed to be pulled from the lab after 1 year and/or residents wished to do additional time for a PhD in addition, we were interested in the scenario in which research became optional and/or was phased out entirely. The macro can also accommodate changes in resident staffing at any level, entry into research at any year, and can be used any time a change occurs in resident complement. It does not address rotations, only total number of residents at each PGY level for that given year. Our macro allows for planning the number of residents that will be able to enter the lab as well as how many will be directly continuing to the clinical practice so the flow of residents will soon converge to/continue to be the ideal flow. The macro allows future planning on the number of interns that will be needed in the following years and allows for planning of expansion. The macro also enables administrators to visualize at what levels and years in the future there will be an excess or shortage of residents at any level depending on the gain or loss of any number of residents over the year. For example, a user may enter the current residents’ names and PGY levels and obtain a matrix with projections for future years. After a few years, if a discrepancy occurs, such as a resident at PGY-3 (SAR1) level leaving and another second year RF resident deciding to stay for an extra year of research (to be RFx), then at that year the user 34
will enter 1 to the loss subfield of SAR1 and again enter 1 to the End_RFx subfield of RF year 2. The formulas at the lower rows in the “Brain Sheet” representing the future years will again show how to reach the ideal flow after these unexpected discrepancies occur. This way the administrators can obtain valuable insights on the affects of such phenomena and will have a chance to adjust their policies accordingly. For instance, according to the parameters of the residency program (minimum CR, maximum CR, minimum I, DV, and steady RF), attrition at a particular level might make it more difficult to rereach the ideal flow, or perhaps letting residents to stay in research more (at RFx) might cause more interns to be recruited to satisfy the minimum CR requirement. We believe these insights are valuable to the administrators of the program and they are achievable via playing with numbers using our “Brain Sheet.” One of the effects to note that the parameters have on the flow of residents is of maximum CR. If the program starts with more residents on research electives than desired, then to decrease research personnel it is important to hire fewer interns and send as many as possible directly to clinical practice; this can only happen if there are additional spaces at the clinic left by the residents finishing research. The limitation on that number of such residents will be determined by maximum CR, therefore this value must be carefully selected. The flexibility we provide in determining the ideal flow according to the user allows for determining the ideal number of residents to have at each PGY level and at each RY. We believe the strength of this model is that we have not only created a way to reach our objective but also provide an easily applicable software tool (an Excel macro) available to the administration of a program that has been made available to the public at: https://www.dropbox. com/s/u6erh09wzvnavls/Resident%20Matrix%20Calculator. xlsm. The formulas are complex, but the Excel macro is easily used. Another note on the parameters would be the comparative relationships they must have to have a feasible model. For example, the desired amount of PGY levels for the clinic should be between minimum CR and maximum CR (minimum CR ⱕ DV ⱕ maximum CR); similarly, minimum I ⱕ DV and minimum I ⱕ maximum CR is recommended unless attrition is very high. Application of these underlying operations in research algorithms is rare in academic surgery. These are the same approaches that Federal Express uses to ensure certain planes are in certain spots at certain times in a rational and cost-effective fashion. We hope the model and formulas given along with the macro and worksheets will provide a useful tool to the administrators of surgery residency programs on making annual staffing and resident allocation decisions, transforming an errorprone manual guess work procedure to an automated and analytically guided decision process. We have used it successfully at Loyola to replace the more typical empiric trial and error approach to overall staffing.
Journal of Surgical Education • Volume 70/Number 1 • January/February 2013
ACKNOWLEDGMENTS
surgery training contribute to graduates’ career choice? Am Surg. 2011;77:907-910.
Is¸ılay Talay-Deg˘irmenci and Casey J. Holmes contributed equally to this work.
4. Baldwin DC Jr, Rowley BD, Daugherty SR, Bay RC. With-
REFERENCES
5. Sue GR, Bucholz EM, Yeo H, et al. The vulnerable stage of
1. Ullrich N, Botelho CA, Hibberd P, Bernstein HH. Re-
search during pediatric residency: predictors and residentdetermined influences. Acad Med. 2003;78:1253-1258. 2. Ellis MC, Dhungel B, Weerasinghe R, Vetto JT, Deveney
K. Trends in research time, fellowship training, and practice patterns among general surgery graduates. J Surg Educ. 2011;68:309-312. 3. Bhattacharya SD, Williams JB, de la Fuente SG, Kuo PC,
Seigler HF. Does protected research time during general
drawal and extended leave during residency training: results of a national survey. Acad Med. 1995;70:1117-1124. dedicated research years of general surgery residency: results of a national survey. Arch Surg. 2011;146:653-658. 6. Robertson CM, Klingensmith ME, Coopersmith CM.
Long-term outcomes of performing a postdoctoral research fellowship during general surgery residency. Ann Surg. 2007;245:516-523.
Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jsurg.2012. 07.003.
Journal of Surgical Education • Volume 70/Number 1 • January/February 2013
35
An Analytic Decision Support Tool for Resident Allocation Is¸ılay Talay-Deg ˘ irmenci, PhD,* Casey J. Holmes, MD,† Paul C. Kuo, MD,† and Otis B. Jennings, PhD‡ *Department of Business Administration, College of Business, Antalya International University, Antalya, Turkey; † Department of Surgery, Loyola University Medical Center, Maywood, Illinois; and ‡Duke University, The Fuqua School of Business, Durham, North Carolina BACKGROUND: Moving residents through an academic residency program is complicated by a number of factors. Across all residency programs the percentage of residents that leave for any reason is between 3.4% and 3.8%.1 There are a number of residents that participate in research. To avoid discrepancies in the number of residents at the all levels, programs must either limit the number of residents that go into the lab, the number that return to clinical duties, or the number of interns to hire. Traditionally this process consists of random selection and trial and error with names on magnetic strips moved around a board. With the matrix that we have developed this process is optimized and aided by a Microsoft Excel macro (Microsoft Corp, Redmond, Washington). METHODS: We suggest that a residency program would have
the same number of residents at each residency stage of clinical practice, as well as a steady number of residents at each research stage. The program consists of 2 phases, in the first phase, an Excel sheet called the “Brain Sheet,” there are simple formulas that we have prepared to determine the number of interns to recruit, residents in the research phase, and residents that advance to the next stage of training. The second phase of the program, the macro, then takes the list of current resident names along with the residency level they are in, and according to the formulas allocates them to the relevant stages for future years, creating a resident matrix. RESULTS: Our macro for resident allocation would maximize the time of residency program administrators by simplifying the movement of residents through the program. It would also provide a tool for planning the number of new interns to recruit and program expansion. CONCLUSIONS: The application of our macro illustrates that analytical techniques can be used to minimize the time
Correspondence: Address inquiries to Paul C. Kuo, MD, MBA, LUMC, Department of Surgery, EMS Building, RM 3244, 2160 South First Ave., Maywood, IL 60153; fax: 708-327-2852; e-mail: [email protected]
spent and avoid the trial and error while planning resident movement in a program. (J Surg 70:31-35. © 2012 Association of Program Directors in Surgery. Published by Elsevier Inc. All rights reserved.) KEY WORDS: operations research, surgery residency, macro,
schedule COMPETENCIES: Patient Care, Practice-Based Learning and Improvement, Systems-Based Practice
INTRODUCTION Academic residency programs fulfill the important mission of training residents as clinical practitioners with specialized expertise. There are a large portion of residency programs that offer a dedicated time for research and others require research experience for completion of program.1 The duration of residency programs in surgery is at least 5 years; those that require research are 7 years. Residents that participate in research traditionally work in the clinical setting for the first 2 years and then leave for their research elective.2,3 These residents spend 2 or more years doing research or earning a PhD, in some cases. Finally, these residents then spend 3 years in clinical training before graduating from the program. Differences in this path may occur, but we suggest our model can be modified according to the specifications of a residency program. Across all residency programs the percentage of residents that leave the program is between 3.4% and 3.8%.4 The number of those who will take a break from clinical duties for a research elective compared with the number of residents that have completed their time in the lab and will return might also vary. The combination of all these factors can lead to a discrepancy in the number of residents at each level of training.5,6 The objective of this report is to provide analytically guided recommendations to the administrators of a surgery residency program for determining the size of the postgraduate year (PGY)-1 class and the number of residents that will continue to the PGY-3 each year. When discrepancies from the ideal number of trainees at each level occur because of attrition and other
Journal of Surgical Education • © 2012 Association of Program Directors in Surgery 1931-7204/$30.00 Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jsurg.2012.07.003
31
factors do occur, this model will allow a residency program to reach the ideal flow in as few years as possible, and at the same time abide by the necessary minimum and maximum number of chief residents (CR) and the minimum number of interns (I). We provide, to our knowledge, a first step to transform this process into an analytically guided method where the current resident names and the number of the residents at each level are entered into an Excel workbook along with the ideal PGY and research year (RY) levels, minimum and maximum CR, minimum I, and the Excel macro yields a resident allocation matrix with names and projections for future years.
METHODS The URL for the Resident Matrix Calculator is https://dl. dropbox.com/u/45447844/Resident%20Matrix%20Calculator. xlsm. Alternatively, the senior author can be contacted. With a significant number of residents taking advantage of research electives or leaving their training program, the spaces left at each PGY level must be filled. Administrators of a program may have to choose the minimum and maximum number of residents required at each PGY level. For example, the last PGY level, the CR, might be required to have a number of residents between predetermined minimum and maximum numbers; and the first PGY level, I, may have a minimum number of spaces required to staff the program. Research is assumed not to be mandatory for our model; therefore it is expected that the number of residents choosing to do research during residency may change from year to year. Financial considerations, such as research budget restraints and the availability of grants also plays a role in the decision of the number or residents allowed to enter the lab and can change annually. Technical considerations, such as equipment and lab space availability are also among the determining factors for the number of spaces open in the research labs. There are 2 decisions the administrators of a residency program make annually: how many interns should be recruited, and how many PGY-2 residents (JARs) must proceed directly to the next clinical training stage, PGY-3 (SAR1). The administrators of the residency programs fortunately have a certain level of control on part of the system dynamics. First of all, they determine the number of I to hire each year. Second, by also considering the financial and regulatory constraints, they can also determine the number of residents that continue clinical practice without participating in research, ie, these residents will proceed to PGY-3 directly from PGY-2 and will not be in the research year (RY) status. Residents at 1 PGY level are expected to advance to the next PGY level the following year; however, the above factors create the desire to avoid large fluctuations among the number of residents at every PGY level. The objective of the formulas is to reach the desired balance for the allocation of the residents in the shortest time possible starting from the current situation. The formulas are created using combinations of the 4 operations and min() and max() built-in functions of Microsoft Excel, and they give the number of interns that need to be recruited and the number of JARs that 32
may proceed to SAR1 each year. In the Excel workbook containing the macro, our formulas are placed in the Excel worksheet called the “Brain Sheet.” In the “Brain Sheet,” the first row of the table is spared to enter the numbers for the current year’s distribution of residents. The user enters the current year and how many residents are at each stage of the residency program. The formulas are in the second row. Thus, after entering the first row, the (already entered) formulas automatically calculate the following year’s distribution and hence shows the number of I to recruit and JARs to proceed to SAR1 the next year, and all the user must do is “pull down” the formulas on the second line and they give the future years’ allocation of residents. If enough rows are created by “pulling down” the formulas, the user will eventually see the desired balance in resident allocation. Our formulas are created so that this will be achieved in the least number of years possible. To increase the versatility we added 3 constraints that the user may enforce; the maximum number of CR, minimum number of CR, and the minimum number of I. If there are no such constraints, then these constraints can be relaxed by entering 0 for the minimum number required and a very large number which will not be reached in reality for the maximum number (eg, 10,000). The program consists of 2 phases; in the first phase, the Excel worksheet called the “Brain Sheet” includes formulas that we prepared to calculate/do the following for future years (Fig. 1): 1. the number of I (PGY-1) to recruit; 2. the number of JARs (PGY-2) that will bypass research and progress to the SAR1 (PGY-3) level; 3. the number of JARs that will enter the research year 1tract; 4. The program will also distribute the correct number of residents to the research year 2, SAR1 (PGY-3), PGY-4 (SAR2), and CR (PGY-5) levels. For residents who would like to advance in their research expertise we also added a research year x (RFx) state where it is possible for residents to enter and stay as long as desired, for example if they want to obtain a PhD. The second phase of the program consists of an Microsoft Excel macro created via Visual Basic integrated design environment, which then takes the list of current resident names along with the residency state they are in, and according to the formulas listed in the “Brain Sheet,” allocates the residents to the relevant stages for future years, creating a resident matrix that shows the annual assignment planning, starting from the current year (Fig. 2). Excel Macro To make full use of the program the user enters the names of the current residents into the “Entry Sheet” and the user enters minimum CR, maximum CR, minimum I, desired value (DV), and steady research fellow (RF) on the “Brain Sheet.” The user then enters the current year and the number of residents at each stage of the current year into the first row of the table in the “Brain Sheet” and then the already entered formulas at the second row automatically give recommended numbers for next
Journal of Surgical Education • Volume 70/Number 1 • January/February 2013
Desired Value 7
Years 2011 2012 2013 2014 2015 2016 2017
Start 7 5 6 7 7 7 7
Maximum CR 8
Interns End_JAR 7 5 6 7 7 7 7
Loss
Minimum Interns 5
Min CR 7
Steady RF 3
For People Doing PhDs JARs RF1 RF2 RFx SAR1 SAR2 CR Start End_SAR1 Loss End_RF1 End_RF2 End_RFx Start End_RF2 Loss End_RFx End_SAR1 Start End_SAR1 Loss End_RFx Start End_SAR1 End_RFx Loss Start End_SAR2 Loss Start End_CR Loss Start End Loss 7 2 5 5 5 6 6 8 8 7 7 7 7 7 3 4 5 5 5 5 0 0 8 8 8 8 7 7 5 2 3 4 4 5 5 0 0 8 8 8 8 8 8 6 3 3 3 3 4 4 0 0 7 7 8 8 8 8 7 4 3 3 3 3 3 0 0 7 7 7 7 8 8 7 4 3 3 3 3 3 0 0 7 7 7 7 7 7 7 4 3 3 3 3 3 0 0 7 7 7 7 7 7
FIGURE 1. An example of the “Brain Sheet.” in which a sample starting resident allocation and parameter set is entered and manipulated. CR, chief resident or PGY-5; Intern, PGY-1; JAR, PGY-2; PGY, postgraduate year; RF1, research fellow first year; RF2, research fellow second year; SAR1, PGY-3; SAR2, PGY-4.
year’s PGY and RY levels. “Pulling down” the formulas at the second row until resident allocation has the desired balance will give the recommended projection to the administrators.
INT
JAR
RF 1
RF 2
SAR1
SAR2
CR
Y ear 2011-2012 .2011.1 .2011.2 .2011.3 .2011.4 .2011.5 .2011.6 .2011.7 .2010.1 .2010.2 .2010.3 .2010.4 .2010.5 .2010.6 .2010.7 .2009.1 .2009.2 .2009.3 .2009.4 .2009.5 .2008.1 .2008.2 .2008.3 .2008.4 .2008.5 .2008.6 .2007.1 .2007.2 .2007.3 .2007.4 .2007.5 .2007.6 .2007.7 .2007.8 .2006.1 .2006.2 .2006.3 .2006.4 .2006.5 .2006.6 .2006.7 .2005.1 .2005.2 .2005.3 .2005.4 .2005.5 .2005.6 .2005.7
INT
Year 2012-2013 2012. 1 2012. 2 2012. 3 2012. 4 2012. 5
JAR
INT
.2011.1 .2011.2 .2011.3 .2011.4 .2011.5 .2011.6 .2011.7 RF 1 .2010.1 .2010.2 .2010.3 .2010.4 .2010.5 RF 2 .2009.1 .2009.2 .2009.3 .2009.4 .2009.5
JAR
SAR 1 .2010.6 .2010.7 .2008.1 .2008.2 .2008.3 .2008.4 .2008.5 .2008.6 SAR2 .2007.1 .2007.2 .2007.3 .2007.4 .2007.5 .2007.6 .2007.7 .2007.8 CR .2006.1 .2006.2 .2006.3 .2006.4 .2006.5 .2006.6 .2006.7
Y ear 2013-2014 2013. 1 2013. 2 2013. 3 2013. 4 2013. 5 2013. 6 2012. 2012. 2012. 2012. 2012.
1 2 3 4 5
INT
JAR
After enabling the macros, when the “Front Sheet” is clicked, the macro automatically creates the resident annual assignment planning matrix using the names from the “Entry Sheet.” The Year 2014-2015 2014. 1 2014. 2 2014. 3 2014. 4 2014. 5 2014. 6 2014. 7 2013. 1 2013. 2 2013. 3 2013. 4 2013. 5 2013. 6 2012. 1 2012. 2 2012. 3
Year 2015-2016 2015. 1 2015. 2 2015. 3 2015. 4 2015. 5 2015. 6 2015. 7 JAR 2014. 1 2014. 2 2014. 3 2014. 4 2014. 5 2014. 6 2014. 7 RF 1 2013. 1 2013. 2 2013. 3
INT
Year 2016-2017 2016. 1 2016. 2 2016. 3 2016. 4 2016. 5 2016. 6 2016. 7 JAR 2015. 1 2015. 2 2015. 3 2015. 4 2015. 5 2015. 6 2015. 7 RF 1 2014. 1 2014. 2 2014. 3
INT
Year 2017-2018 2017. 1 2017. 2 2017. 3 2017. 4 2017. 5 2017. 6 2017. 7 JAR 2016. 1 2016. 2 2016. 3 2016. 4 2016. 5 2016. 6 2016. 7 RF 1 2015. 1 2015. 2 2015. 3
INT
RF 1 .2011.1 . 2011. 2 . 2011. 3 . 2011. 4
RF 1
RF 2 .2010.1 . 2010. 2 . 2010. 3 . 2010. 4 . 2010. 5
RF 2 .2011.1 . 2011. 2 . 2011. 3 . 2011. 4
RF 2
SAR 1 .2011.5 . 2011. 6 . 2011. 7 . 2009. 1 . 2009. 2 . 2009. 3 . 2009. 4 . 2009. 5 SAR2 .2010.6 . 2010. 7 . 2008. 1 . 2008. 2 . 2008. 3 . 2008. 4 . 2008. 5 .2008.6 CR .2007.1 .2007. 2 .2007. 3 .2007. 4 .2007. 5 .2007. 6 .2007. 7 .2007.8
SAR 1 2012. 4 2012. 5 . 2010. 1 . 2010. 2 . 2010. 3 . 2010. 4 . 2010. 5
SAR 1 2013. 4 2013. 5 2013. 6 . 2011.1 . 2011.2 . 2011.3 . 2011.4
SAR 1 2014. 2014. 2014. 2014. 2012. 2012. 2012.
4 5 6 7 1 2 3
SAR 1 2015. 2015. 2015. 2015. 2013. 2013. 2013.
4 5 6 7 1 2 3
SAR2 .2011.5 . 2011. 6 . 2011. 7 . 2009. 1 . 2009. 2 . 2009. 3 . 2009. 4 .2009.5 CR .2010.6 . 2010. 7 . 2008. 1 . 2008. 2 . 2008. 3 . 2008. 4 . 2008. 5 .2008.6
SAR2 2012. 4 2012. 5 . 2010.1 . 2010.2 . 2010.3 . 2010.4 . 2010.5
SAR2 2013. 4 2013. 5 2013. 6 .2011.1 .2011.2 .2011.3 .2011.4
SAR2 2014. 2014. 2014. 2014. 2012. 2012. 2012.
4 5 6 7 1 2 3
CR
CR
CR
2012. 1 2012. 2 2012. 3
.2011.5 .2011.6 .2011.7 .2009.1 .2009.2 .2009.3 .2009.4 .2009.5
RF 2
2013. 1 2013. 2 2013. 3
2012. 4 2012. 5 .2010. 1 .2010. 2 .2010. 3 .2010. 4 .2010. 5
RF 2
2014. 1 2014. 2 2014. 3
2013. 4 2013. 5 2013. 6 .2011. 1 .2011. 2 .2011. 3 .2011. 4
FIGURE 2. An example of the “Front Sheet” or the matrix of residents and how they are predicted to progress through the training program based on the formulas contained in the “Brain sheet” and names entered on the “Entry Sheet.” CR, chief resident or PGY-5; Intern, PGY-1; JAR, PGY-2; PGY, postgraduate year; RF1, research fellow first year; RF2, research fellow second year; SAR1, PGY-3; SAR2, PGY-4. Journal of Surgical Education • Volume 70/Number 1 • January/February 2013
33
macro created for demonstrational purposes of this project by default starts placing the JARs to first year RF from the top of the list until the desired amount dictated by the formula is reached, then places the rest of the JARs to SAR1 for next year. When resident names are entered, it could be more suitable to list those who are more likely to choose research electives at the top of their respective list corresponding to their residency stage.
DISCUSSION Moving residents through an academic residency program is a complicated process and takes a considerable amount of time and resources. All residency programs deal with attrition and many deal with residents taking time from clinic duties to participate in research. These issues are further complicated by expansion of the program, and varying numbers of residents that leave for the lab versus the number returning. The combination of these factors leads to a discrepancy in the number of residents at each level of training. To avoid discrepancies in the number of CR, programs either limit the number of residents that go into the lab or the number that directly continues to clinical duties. The program must also consider the size of the PGY-1 class and the number of residents that have left the program. Traditionally, the process of selecting residents to move into and out of the lab consists of random trial and error. With the macro that we have developed this process is transformed from error-prone guess work to an analytically guided decision process. The macro actually was designed for a program with mandatory research years of a prescribed duration. Situations arose in which residents needed to be pulled from the lab after 1 year and/or residents wished to do additional time for a PhD in addition, we were interested in the scenario in which research became optional and/or was phased out entirely. The macro can also accommodate changes in resident staffing at any level, entry into research at any year, and can be used any time a change occurs in resident complement. It does not address rotations, only total number of residents at each PGY level for that given year. Our macro allows for planning the number of residents that will be able to enter the lab as well as how many will be directly continuing to the clinical practice so the flow of residents will soon converge to/continue to be the ideal flow. The macro allows future planning on the number of interns that will be needed in the following years and allows for planning of expansion. The macro also enables administrators to visualize at what levels and years in the future there will be an excess or shortage of residents at any level depending on the gain or loss of any number of residents over the year. For example, a user may enter the current residents’ names and PGY levels and obtain a matrix with projections for future years. After a few years, if a discrepancy occurs, such as a resident at PGY-3 (SAR1) level leaving and another second year RF resident deciding to stay for an extra year of research (to be RFx), then at that year the user 34
will enter 1 to the loss subfield of SAR1 and again enter 1 to the End_RFx subfield of RF year 2. The formulas at the lower rows in the “Brain Sheet” representing the future years will again show how to reach the ideal flow after these unexpected discrepancies occur. This way the administrators can obtain valuable insights on the affects of such phenomena and will have a chance to adjust their policies accordingly. For instance, according to the parameters of the residency program (minimum CR, maximum CR, minimum I, DV, and steady RF), attrition at a particular level might make it more difficult to rereach the ideal flow, or perhaps letting residents to stay in research more (at RFx) might cause more interns to be recruited to satisfy the minimum CR requirement. We believe these insights are valuable to the administrators of the program and they are achievable via playing with numbers using our “Brain Sheet.” One of the effects to note that the parameters have on the flow of residents is of maximum CR. If the program starts with more residents on research electives than desired, then to decrease research personnel it is important to hire fewer interns and send as many as possible directly to clinical practice; this can only happen if there are additional spaces at the clinic left by the residents finishing research. The limitation on that number of such residents will be determined by maximum CR, therefore this value must be carefully selected. The flexibility we provide in determining the ideal flow according to the user allows for determining the ideal number of residents to have at each PGY level and at each RY. We believe the strength of this model is that we have not only created a way to reach our objective but also provide an easily applicable software tool (an Excel macro) available to the administration of a program that has been made available to the public at: https://www.dropbox. com/s/u6erh09wzvnavls/Resident%20Matrix%20Calculator. xlsm. The formulas are complex, but the Excel macro is easily used. Another note on the parameters would be the comparative relationships they must have to have a feasible model. For example, the desired amount of PGY levels for the clinic should be between minimum CR and maximum CR (minimum CR ⱕ DV ⱕ maximum CR); similarly, minimum I ⱕ DV and minimum I ⱕ maximum CR is recommended unless attrition is very high. Application of these underlying operations in research algorithms is rare in academic surgery. These are the same approaches that Federal Express uses to ensure certain planes are in certain spots at certain times in a rational and cost-effective fashion. We hope the model and formulas given along with the macro and worksheets will provide a useful tool to the administrators of surgery residency programs on making annual staffing and resident allocation decisions, transforming an errorprone manual guess work procedure to an automated and analytically guided decision process. We have used it successfully at Loyola to replace the more typical empiric trial and error approach to overall staffing.
Journal of Surgical Education • Volume 70/Number 1 • January/February 2013
ACKNOWLEDGMENTS
surgery training contribute to graduates’ career choice? Am Surg. 2011;77:907-910.
Is¸ılay Talay-Deg˘irmenci and Casey J. Holmes contributed equally to this work.
4. Baldwin DC Jr, Rowley BD, Daugherty SR, Bay RC. With-
REFERENCES
5. Sue GR, Bucholz EM, Yeo H, et al. The vulnerable stage of
1. Ullrich N, Botelho CA, Hibberd P, Bernstein HH. Re-
search during pediatric residency: predictors and residentdetermined influences. Acad Med. 2003;78:1253-1258. 2. Ellis MC, Dhungel B, Weerasinghe R, Vetto JT, Deveney
K. Trends in research time, fellowship training, and practice patterns among general surgery graduates. J Surg Educ. 2011;68:309-312. 3. Bhattacharya SD, Williams JB, de la Fuente SG, Kuo PC,
Seigler HF. Does protected research time during general
drawal and extended leave during residency training: results of a national survey. Acad Med. 1995;70:1117-1124. dedicated research years of general surgery residency: results of a national survey. Arch Surg. 2011;146:653-658. 6. Robertson CM, Klingensmith ME, Coopersmith CM.
Long-term outcomes of performing a postdoctoral research fellowship during general surgery residency. Ann Surg. 2007;245:516-523.
Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jsurg.2012. 07.003.
Journal of Surgical Education • Volume 70/Number 1 • January/February 2013
35