ORIGINAL REPORTS
Productivity Change of Surgeons in an Academic Year Yoshinori Nakata, MD,* Yuichi Watanabe, MS,* Hiroshi Otake, MD,† Toshihito Nakamura, PhD,‡ Giichiro Oiso, MD,§ and Tomohiro Sawa, PhDǁ Teikyo University Graduate School of Public Health, Tokyo, Japan; †Department of Anesthesia, Showa University School of Medicine, Tokyo, Japan; ‡Department of Health Policy for Aged Society, Chiba University Hospital, Chiba, Japan; §Hamamatsu University School of Medicine, Hamamatsu, Japan; and ǁ Teikyo University Medical Information and System Research Center, Tokyo, Japan *
OBJECTIVE: The goal of this study was to calculate total
KEY WORDS: Malmquist index, efficiency change, tech-
factor productivity of surgeons in an academic year and to evaluate the effect of surgical trainees on their productivity.
nical change, surgical training COMPETENCIES: Patient Care, Practice-Based Learning
and Improvement, Systems-Based Practice
STUDY DESIGN: We analyzed all the surgical procedures
performed from April 1 through September 30, 2013 in the Teikyo University Hospital. The nonradial and nonoriented Malmquist model under the variable returns-to-scale assumptions was employed. A decision-making unit is defined as a surgeon with the highest academic rank in the surgery. Inputs were defined as the number of physicians who assisted in surgery, and the time of surgical operation from skin incision to skin closure. The output was defined as the surgical fee for each surgery. April is the beginning month of a new academic year in Japan, and we divided the study period into April to June and July to September 2013. We computed each surgeon’s Malmquist index, efficiency change, and technical change. RESULTS: We analyzed 2789 surgical procedures that were performed by 105 surgeons. The Malmquist index of all surgeons was significantly greater than 1 (p ¼ 0.0033). The technical change was significantly greater than 1 (p o 0.0001). However, the efficiency change was not statistically significantly different from 1 (p ¼ 0.1817). CONCLUSIONS: The surgeons are less productive in the
beginning months of a new academic year. The main factor of this productivity loss is considered to be surgical training. C 2014 Association of Program ( J Surg 72:128-134. J Directors in Surgery. Published by Elsevier Inc. All rights reserved.)
This work was supported by JSPS KAKENHI Grant no. 25670337, which was given to Dr Yoshinori Nakata. Correspondence: Inquiries to Yoshinori Nakata, MD, MBA, Teikyo University Graduate School of Public Health, 2-11-1 Kaga, Itabashi-ku, Tokyo 173-8605, Japan; fax: (339) 63-2687; e-mail: [email protected]
128
INTRODUCTION Operating room efficiency is an important concern in most hospitals today.1 The most significant part of the operating room efficiency is considered to be surgical procedures because they usually occupy the longest time portion of the operating room time. When a new academic year starts, a number of new interns, residents, fellows, and attending physicians begin to work in a teaching hospital. It has been suggested that new trainees adversely affect the productivity in the operating rooms of a university hospital.2 In fact, it is demonstrated that surgical trainees’ participation in surgical procedures is associated with an increase in total operative time.3 Although those studies may be clinically significant, they only represent partial productivity measures, which evaluate only 1 factor (input) and 1 product (output), while assuming other variables as constant. A partial productivity measure may impute productivity changes to 1 input (or 1 output) that are really attributable to some other input (or output).4 However, there are no studies that have evaluated total factor productivity of surgeons that combines all inputs and all outputs. The Malmquist productivity index (Malmquist index, MI) represents total factor productivity change of a decision-making unit (DMU) between 2 periods under dynamic situation and is an example of in comparative statics analysis.4 It is based on data envelopment analysis (DEA), which evaluates relative efficiency of DMUs against the efficient frontier under static conditions in a single period. By comparing DEA results between 2 time periods, MI can divide productivity change into 2 components, one measuring efficiency change (EC) and the other measuring
Journal of Surgical Education & 2014 Association of Program Directors in Surgery. Published by 1931-7204/$30.00 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jsurg.2014.06.022
technical change (TC).5 MI is defined as the product of EC and TC terms. The EC term relates to the degree to which a DMU improves or worsens its efficiency, whereas the TC term reflects the change in the efficient frontiers between the 2 time periods. If productivity change of a DMU is compared between period 1 and period 2, they are mathematically defined as follows4: EC ¼ Efficiency of a DMU in period 2 with respect to period 2 frontier = Efficiency of a DMU in period 1 with respect to period 1 frontier TC ¼
Efficiency of a DMU in period 1 with respect to period 1 frontier = Efficiency of a DMU in period 1 with respect to period 2 frontier Efficiency of a DMU in period 2 with respect to period 1 frontier = Efficiency of a DMU in period 2 with respect to period 2 frontier 1=2
MI ¼ EC TC for example, we choose a 2-input, 1-output model that we will use in this study (Fig.). EC is defined as follows: EC ¼
OA=OB OE =OG
TC is defined as follows: OE=OG OC =OB 1=2 OE OC 1=2 ¼ TC ¼ OF =OG OA=OB OF OA thus, combining EC with TC, we get OC =OB OA=OB 1=2 MI ¼ EC TC ¼ OE=OG OF =OG
TABLE 1. Technical Terminology for Productivity Study Factor
The input used to produce goods and services Input Same as factor or resource Product The goods and services produced from the inputs Output Same as product or goods and services Total factor productivity Productivity that includes all the outputs produced and accounts for all of the inputs used to produce these outputs Decision-making unit The entity that is regarded as (DMU) responsible for converting inputs into outputs Data envelopment A frontier analysis method that analysis (DEA) evaluates relative efficiency of DMUs Efficient frontier A combination of minimum inputs that produces a certain amount of outputs, or a combination of maximum outputs that are produced from a certain amount of inputs Nonradial, nonoriented A variant of Malmquist model that Malmquist model takes account of all existing slacks (input excesses and output shortfalls) Variable returns to scale A production function that assumes that outputs increase variably when inputs increase proportionally Comparative statics An analysis that compares analysis 2 different economic outcomes, before and after a change
where OA, OB, OC, OE, OF, and OG are defined as the linear distances from the origin to the respective points. There have been a number of studies that applied DEA to surgery, and they provided clinically significant results.6-8 Although there has never been a study that applied a DEAbased MI model to surgery, it is considered appropriate to do so because MI is a nonparametric DEA model under time-dependent situations.4 Moreover, the MIs have been used to assess productivity change in a variety of sectors, such as agriculture, airlines, banking, electric utilities, insurance companies, and public sectors.5 Technical terminology used in efficiency and productivity study is summarized in Table 1. The goal of this study was to compute total factor productivity of surgeons in an academic year and to evaluate the effect of surgical trainees on their productivity change.
METHODS FIGURE. The Malmquist index in a 2-input, 1-output model. A DMU is in G in period 1 and moves to B in period 2. P1 indicates the efficient frontier in period 1. P2 indicates the efficient frontier in period 2.
The Teikyo University institutional review board approved our study. Anonymity of the data was strictly maintained by de-identification by the research team.
Journal of Surgical Education Volume 72/Number 1 January/February 2015
129
Data Teikyo University Hospital is located in the metropolitan Tokyo, Japan, serving a population of 1,000,000. It has 1152 beds and has a surgical volume of approximately 9000 cases annually. It has 13 surgical specialty departments. We collected data from surgical records in the Teikyo University Hospital electronic medical record system. One of the authors (Y.N.) looked up in person all the surgical procedures performed from April 1 through September 30, 2013 in the main operating rooms of Teikyo University Hospital and extracted the necessary information from the electronic medical record system. Exclusion criteria for surgery were as follows. First, surgical procedures performed under local anesthesia by surgeons were excluded to equalize resource utilization other than the inputs that we defined later. Oral, dermatologic, and ophthalmologic surgical procedures were excluded because most of their cases were minor surgeries performed under local anesthesia without anesthesiologists’ involvement, and those under general anesthesia do not represent the activity of their surgeons. Second, the surgical procedures were excluded if the patients die within a month after surgery to maintain the quality of surgery. Third, the surgical procedures that were not reimbursed under the current surgical payment system were excluded. Fourth, the surgical procedures were excluded if their records were incomplete for any reason. Analysis Framework We employed the nonradial and nonoriented Malmquist model under the variable returns-to-scale assumptions, which was particularly relevant because of its ability to employ multiple inputs and outputs simultaneously.9 In this analysis, we focused on the surgeons’ activity and their clinical decision. We defined in this study the DMU as a surgeon with the highest academic rank that scrubbed in the surgery. All the inputs and outputs are under the control of a DMU. Inputs were defined as (1) the number of physicians who assisted in surgery (assistants) and (2) the time of surgical operation from skin incision to skin closure (surgical time). The output was defined as the surgical fee for each surgery. It is classified as K000-K915 in the Japanese surgical fee schedule and is called “K codes.” Each surgical procedure is assigned to one of the K codes, which correspond with surgical fees. The fee is identical regardless of who (a senior surgeon or a surgical trainee) performs surgery as long as they have medical licensure, how many assistants they use, or how long it takes to complete surgery.10 The additional reimbursements for expensive surgical devices, such as automatic suture devices or imaging navigation devices, were excluded. Other fees for blood transfusion, medications, special insurance medical materials, and anesthesia were also excluded. The monetary values 130
of surgical fees were originally expressed in the Japanese yen, and were converted to US $ at $1 ¼ 100 yen to facilitate understanding by international readers. April is the beginning month of a new academic year in Japan. A number of new attending surgeons and trainees start working on April 1. They need some time to accustom themselves to a new operating room environment and new surgical instruments. In our hospital, new interns start working in May after a month’s orientation sessions. Therefore, we divided the study period into 2 periods: April to June 2013 (period 1) and July to September 2013 (period 2). We added all the inputs and outputs of the surgical procedures for each DMU during these study periods and computed his/her MI, EC, and TC using DEA-Solver-Pro Software (Saitech, Inc., Tokyo, Japan).4 MI 4 1 indicates progress in total factor productivity of the DMU from period 1 to 2, whereas MI ¼ 1 and MI o 1, respectively, indicate the status quo and deterioration in the total factor productivity. Similarly, a numerical value for the EC and the TC measure of greater than 1 implies that there is efficiency and technical progress, respectively. All the surgeons in the sample were given an MI, EC, and TC for each.11 All the surgeons analyzed were employees of Teikyo University and were salaried according to their ranks and experiences without any monetary incentives to increase surgical volume or productivity. The hospital charges surgeons’ surgical fees to Health Insurance Claims Review & Reimbursement Services, and the reimbursement becomes the revenue of the hospital. It pays surgeons their salary from this revenue. The surgeons belong to one of the following 10 surgical specialty departments: cardiovascular surgery, emergency surgery, general surgery, neurosurgery, obstetrics and gynecology, orthopedics, otorhinolaryngology, plastic surgery, thoracic surgery, and urology. We compiled their MIs, ECs, and TCs in their surgical specialties and calculated their means and standard deviations. We excluded from our analysis the surgeons who performed surgery in only 1 of these 2 periods. Statistical Analysis We used Excel Statistics 2008 Software (SSRI Co., Ltd., Tokyo, Japan) for our statistical analysis. Demographic data were analyzed within each study period with the 1-way analysis of variance. Multiple comparison tests were performed using the Bonferroni t tests.12 We also compared demographic data between both the study periods using the paired t tests. We compared the MIs, ECs, and TCs of all surgeons and those of each surgical specialty against 1 using the Student t test. The MIs, ECs, and TCs of each surgical specialty were compared against 1 using the Student t test. We also compared the MI, EC, and TC of each surgical specialty against other specialties using the 1-way analysis of variance.
Journal of Surgical Education Volume 72/Number 1 January/February 2015
TABLE 2. Demographic Data in Period 1 (April to June 2013) Specialty Cardiovascular surgery Emergency surgery General surgery Neurosurgery Obstetrics and gynecology Orthopedics Otorhinolaryngology Plastic surgery Thoracic surgery Urology All surgical procedures
DMUs 5 12 18 5 9 24 10 9 4 9 105
Cases 102 169 236 63 159 277 101 88 59 101 1355
Assistants Per Case 1.25 1.71 1.81 1.42 1.78 1.73 0.83 1.15 1.06 1.43 1.52
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
Time Per Case (min)
0.47 0.63 0.76 0.42 0.37 0.72 0.48║ 0.60 0.26 0.30 0.66
214 131 201 165 83 119 96 120 89 115 135
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
*
65 41 96‡ 68 35 46 41 58 25 60 70
Fee Per Case (US $) 6732 2544 3564 5775 2335 2660 1435 2100 5613 2082 3014
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
3697† 1364 1802 4194§ 977 1634 1313 1177 1847¶ 943 2200
Characteristics of each surgical specialty. Assistants per case, time pre case, and fee per case are expressed in mean ⫾ standard deviation. *The value is significantly different from obstetrics and gynecology and otorhinolaryngology (p o 0.05). † The value is significantly different from emergency surgery, general surgery, obstetrics and gynecology, orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). ‡ The value is significantly different from obstetrics and gynecology, orthopedics, otorhinolaryngology, thoracic surgery, and urology (p o 0.05). § The value is significantly different from obstetrics and gynecology, orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). The value is significantly different from emergency surgery, general surgery, obstetrics and gynecology, and orthopedics (p o 0.05). ¶ The value is significantly different from otorhinolaryngology (p o 0.05).
Multiple comparison tests were performed using the Bonferroni t tests. A p o 0.05 was considered statistically significant.12
RESULTS We analyzed 2789 surgical procedures performed by 105 surgeons (DMUs) during both the study periods. The demographic data of each surgical specialty during both the study periods were shown in Tables 2 and 3. There were significant differences among surgical specialties in assistants per case, surgical time per case, and surgical fee per case (p o 0.05). Between both the study periods, there were no significant differences in assistants per case, surgical
time per case, and surgical fee per case within each surgical specialty except assistants per case in emergency surgery (p ¼ 0.0012). The MIs, ECs, and TCs were shown in Table 4. The MI of all surgeons was significantly greater than 1 (p ¼ 0.0033), which demonstrated that the total factor productivity improved significantly from period 1 to period 2. The TC of all surgeons was significantly greater than 1 (p o 0.0001). However, the EC of all surgeons was not significantly different from 1 (p ¼ 0.1817). The natural logarithms of the MI, TC, and EC allow us to interpret these results as percent change.13 On average, the surgeons were 27.8% more productive in period 2 than in period 1. This productivity growth was decomposed into 3.0% increase in EC and 24.8% increase in TC.
TABLE 3. Demographic Data in Period 2 (July to September 2013) Specialty
Cases
Cardiovascular surgery Emergency surgery General surgery Neurosurgery Obstetrics and gynecology Orthopedics Otorhinolaryngology Plastic surgery Thoracic surgery Urology All surgical procedures
113 179 249 65 182 264 119 83 63 117 1434
Assistants Per Case 1.07 1.49 1.84 1.37 1.96 1.92 0.87 1.14 1.49 1.42 1.57
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
Time Per Case (min)
0.90 0.53† 0.61 0.46 0.24 0.73 0.50║ 0.59 0.24 0.68 0.69
186 121 186 180 93 104 96 122 107 119 129
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
92 50 67‡ 64 27 53 46 57 11 61 64
Fee Per Case (US $) 6125 3040 3665 6473 3182 2247 1427 1921 7259 1974 3108
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
4702* 2513 1414 3103§ 789 1709 1238 1425 563¶ 947 2377
Characteristics of each surgical specialty. Assistants per case, time per case, and fee per case are expressed in mean ⫾ standard deviation. *The value is significantly different from orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). † The value is significantly different from the corresponding value in period 1 (p ¼ 0.0012). ‡ The value is significantly different from obstetrics and gynecology, orthopedics, and otorhinolaryngology (p o 0.05). § The value is significantly different from emergency surgery, orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). The value is significantly different from general surgery, obstetrics and gynecology, and orthopedics (p o 0.05). ¶ The value is significantly different from emergency surgery, general surgery, obstetrics and gynecology, orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). Journal of Surgical Education Volume 72/Number 1 January/February 2015
131
TABLE 4. Malmquist Index, Efficiency Change, and Technical Change of All Surgeons and of Each Surgical Specialty Specialty All surgeons Cardiovascular surgery Emergency surgery General surgery Neurosurgery Obstetrics and Gynecology Orthopedics Otorhinolaryngology Plastic surgery Thoracic surgery Urology
Malmquist Index 1.32 1.28 1.17 1.11 1.27 1.40 1.40 2.23 0.93 1.11 1.17
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
*
1.17 0.51 0.32* 0.34 0.39 0.78 1.65 2.49 0.38 0.24 0.58
Efficiency Change 1.03 1.22 0.98 0.97 1.12 1.01 1.09 1.81 0.76 1.03 0.89
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
0.87 0.41 0.24 0.31 0.19 0.23 1.06 2.14 0.42 0.11 0.24
Technical Change 1.28 1.05 1.20 1.16 1.12 1.45 1.45 1.33 1.30 1.07 1.28
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
0.58* 0.18 0.17* 0.20* 0.20 0.89 1.01* 0.33* 0.25* 0.13 0.30*
*The value is significantly greater than 1 (p o 0.05).
The MI of emergency surgeons was significantly greater than 1 (p ¼ 0.043). The TCs of emergency (p ¼ 0.001), general (p ¼ 0.002), orthopedic (p ¼ 0.020), otorhinolaryngologic (p ¼ 0.005), plastic (p ¼ 0.003), and urologic surgeons (p ¼ 0.013) were significantly greater than 1. The difference in MIs, ECs, and TCs was not statistically significant among surgical specialties (p ¼ 0.5277, 0.4124, and 0.7959, respectively). Using the Bonferroni t tests, we did not find any significant difference in MIs, ECs, and TCs among surgical specialties, either.
DISCUSSION We demonstrated that the surgeons’ MI in the operating rooms was significantly greater than 1. This demonstrates that the surgeons’ productivity improved significantly from period 1 to period 2. This is the first study that evaluated the surgeons’ productivity using the Malmquist model. The surgeons were less productive in the beginning months of a new academic year than in the subsequent months. The 3-month study period was so short that any confounding factors were minimized. This suggests that the newly assigned trainees may be main factors of this productivity loss. The source of productivity growth is due to TC, not due to EC because TC was significantly greater than 1 and EC was not different from 1. This means that productivity growth is because surgical technique improved from period 1 to period 2.4 This finding is consistent with an initial progress of surgical technique of surgical trainees. However, their relative efficiency did not change from period 1 to period 2.4 Surgical training reduces surgeons’ productivity in 2 ways. First, it takes longer to complete surgery with participation of newly assigned trainees than those accustomed to their surgery.3 Second, the surgeons use more assistants as a means of education. The physicians who assist in surgery are usually surgical trainees in teaching hospitals. Those 2 variables were not statistically different between both the periods except the number of assistants in 132
emergency surgery when they were compared separately. However, the productivity demonstrated significant changes when the variables were evaluated simultaneously as demonstrated in total factor productivity measures. These 2 factors also suggest that the surgical training may negatively affect productivity of surgeons in the beginning of a new academic year. The current system for the reimbursement of surgical fees to hospitals in Japan is extremely vague, providing only total prices, adjusted to cover costs for each surgery with no explanation of component costs.14 In addition, the reimbursement system does not consider surgical training cost at all; the reimbursement is identical regardless of whether they are teaching hospitals or private hospitals.10 This reimbursement system discourages surgical training. Surgical training significantly reduces surgeons’ productivity, as demonstrated in this study. Therefore, the reimbursement should take surgical training cost into account and compensate the hospitals for the productivity loss of surgeons at least for 3 months in the beginning of a new academic year. We did not find any significant difference in MIs among surgical specialties. All the surgical specialties uniformly improved their productivity from the beginning months to the subsequent months. This suggests that the productivity loss of surgeons due to surgical training is common among all surgical specialties. Japanese hospitals should be compensated for the productivity loss of their surgeons of all surgical specialties. For example, the hospitals should be reimbursed with 27.8% more surgical fees for the first 3 months of an academic year to compensate for surgeons’ productivity loss because surgeons were 27.8% more productive on average in period 2 than in period 1 as our results demonstrated. As demonstrated in Tables 2 and 3, surgery is widely different in resource utilization and surgical fees among surgical specialties. These characteristics suggest that the Malmquist model that we employed was appropriate because of its ability to take simultaneously multiple different inputs and outputs into account.4 The simple comparison of surgical time, number of assistants, or surgical fees
Journal of Surgical Education Volume 72/Number 1 January/February 2015
does not reflect the heterogeneity of surgery among different surgical specialties. Its evaluation tends to be limited within a few similar types of common surgical procedures.3 Our method is better in aggregating heterogeneous surgical procedures and simultaneously comparing different surgical procedures by taking all inputs and outputs into account. We selected in this study only 2 inputs and a single output for surgery, although there are a number of other inputs and outputs of surgery. For example, anesthesia time is not included in our analysis because it is separately billed and reimbursed. Turnover time is not under surgeons’ control. We did not take into consideration technical difficulty of surgery, training period required to achieve clinical skills, or value of life for patients created by surgery in our analysis. For surgeons’ activity and their clinical decision, although surgical time and the number of assistants are objective measurements, technical difficulty and the value of life are not.14 The problem of technical difficulty regarding the patient’s condition requires an adjustment by patient or patient group, which makes interspecialty comparison impossible. However, the technical difficulty regarding the patient’s condition is not under the control of surgeons (DMUs). Moreover, the technical difficulty is expected to correlate positively with surgical time because the higher the technical difficulty, the longer the surgery. The training period required also positively correlates with technical difficulty. The correlated variables would only decrease the discriminatory power of the model without adding useful information.15 It is acceptable to eliminate one of the factors that are correlated with each other.16 Therefore, we selected these variables, although we know that they do not reflect all aspects of surgery. The effect of more surgical assistants is not necessarily in the operating room, but in other parts of the hospital. If they are tied up in the operating room, they cannot do any other work in the ward or outpatient clinic. If the third and fourth assistants do not help surgeons, they should instead work in the other part of the hospital. Therefore, we assumed from the hospital’s viewpoint that the fewer the assistants, the more productive the surgeons were. There are some limitations in our study. First, this is a study conducted in a single large teaching hospital in Tokyo, Japan. Our surgeons may not represent all the surgeons in Japan. However, this is the first study to evaluate the surgeons’ total factor productivity using the actual data in the Malmquist model. Second, this study is based on the Japanese surgical reimbursement system. Our results may be invalid under a reimbursement system that is different from the Japanese system. Third, we compared the first 3 months of a new academic year to the subsequent 3 months; therefore, the productivity improvement may be due to other factors, such as seasonal case-mix variation. However, as seen in the demographic data, there were no significant differences between both the study periods in inputs and outputs within each surgical specialty, except
assistants per case in emergency surgery. Moreover, each surgeon performs only 5 to 10 types of surgical procedures at most, and his/her operative activity is fairly constant because we studied 2 consecutive 3-month periods. It is expected that there is no significant difference in case-mix variation. Furthermore, 3-month periods are short enough to minimize any other confounding factors, such as change in surgeons, surgical practice patterns, nursing practice, and reimbursement system. Therefore, 3-month study periods are appropriate and the main factor of this productivity change is speculated to be due to surgical training. Fourth, we simply considered the number of assistants without taking their experience into account. It is obvious that a surgeon of lesser rank or a newly appointed senior resident is not equivalent to junior surgical trainees in assisting surgery. An unequal distribution of residents with different experience is also likely. On services like cardiovascular, thoracic, and neurosurgery, more senior trainees need to be present, and it is unlikely that a junior trainee would get to do anything technically, or perhaps even scrub. However, the detailed data were unavailable.
CONCLUSION We demonstrated by the Malmquist model that the surgeons are less productive in the beginning months of a new academic year. The main factor of this productivity loss may be due to surgical training.
REFERENCES 1. Eappen S, Flanagan H, Bhattacharyya N. Introduction
of anesthesia resident trainees to the operating room does not lead to changes in anesthesia-controlled times for efficiency measures. Anesthesiology. 2004;101(5): 1210-1214. 2. Posner KL, Freund PR. Resident training level and
quality of anesthetic care in a university hospital. Anesth Analg. 2004;98(2):437-442. 3. Papandria D, Rhee D, Ortega G, et al. Assessing
trainee impact on operative time for common general surgical procedures in ACS-NSQIP. J Surg Educ. 2012;69(2):149-155. 4. Cooper WW, Seiford LM, Tone K. Data Envelop-
ment Analysis: A Comprehensive Text With Models, Applications, References and DEA-Solver Software. 2nd ed New York: Springer, 2007. 328-347. 5. Coelli TJ, Rao DSP, O’Donnell CJ, Battese GE. An
Introduction to Efficiency and Productivity Analysis. 2nd ed New York: Springer, 2005. 289-310, 76-77.
Journal of Surgical Education Volume 72/Number 1 January/February 2015
133
6. Dexter F, O’Neill L. Data envelopment analysis to
11. Hollingsworth B, Peacock SJ. Efficiency Measurement in
determine by how much hospitals can increase elective inpatient surgical workload for each specialty. Anesth Analg. 2004;99(5):1492-1500.
Health and Health Care. Oxon: Routledge, 2008. 37-39.
7. Basson MD, Butler T. Evaluation of operating room
suite efficiency in the Veterans Health Administration system by using data-envelopment analysis. Am J Surg. 2006;192(2):649-656. 8. O’Neill L, Dexter F. Tactical increase in operating
room block time based on financial data and market growth from data envelopment analysis. Anesth Analg. 2007;104(2):355-368. 9. Thanassoulis E, Portela MCS, Despic O. Data
envelopment analysis: the mathematical programming approach to efficiency analysis. In: Fried HO, Lovell CAK, Schmidt SS, eds. The Measurement of Productive Efficiency and Productivity Growth. New York: Oxford University Press, 2008. 251-420. 10. Social Insurance Institute (Shakaihoken Kenkyuusho).
Interpretation of medical reimbursement scales, April 2012 (Ikatensuuhyou no Kaishaku). 2012: 622-780 [in Japanese].
134
12. Glantz SA, Slinker BK. Primer of Applied Regression
and Analysis of Variance. New York: McGraw-Hill; 1990. 272-313. 13. Fare R, Grosskopf S, Lindgren B, Roos P. Productivity
developments in Swedish hospitals: a Malmquist output index approach. In: Charnes A, Cooper W, Lewin AY, Seiford LM, eds. Data Envelopment Analysis: Theory, Methodology and Applications. New York: Springer, 1994. 253-272. 14. Hayashida K, Imanaka Y. Inequity in the price of
physician activity across surgical procedures. Health Policy. 2005;74:24-38. 15. Paradi JC, Yang Z, Zhu H. Assessing bank and bank
branch performance; modeling considerations and approaches. In: Cooper WW, Seiford LM, Zhu J, eds. Handbook on Data Envelopment Analysis. 2nd ed. New York: Springer, 2011. 315-362. 16. Ozcan YA. Health Care Benchmarking and Perform-
ance Evaluation: An Assessment Using Data Envelopment Analysis. New York: Springer; 2008.
Journal of Surgical Education Volume 72/Number 1 January/February 2015
Productivity Change of Surgeons in an Academic Year Yoshinori Nakata, MD,* Yuichi Watanabe, MS,* Hiroshi Otake, MD,† Toshihito Nakamura, PhD,‡ Giichiro Oiso, MD,§ and Tomohiro Sawa, PhDǁ Teikyo University Graduate School of Public Health, Tokyo, Japan; †Department of Anesthesia, Showa University School of Medicine, Tokyo, Japan; ‡Department of Health Policy for Aged Society, Chiba University Hospital, Chiba, Japan; §Hamamatsu University School of Medicine, Hamamatsu, Japan; and ǁ Teikyo University Medical Information and System Research Center, Tokyo, Japan *
OBJECTIVE: The goal of this study was to calculate total
KEY WORDS: Malmquist index, efficiency change, tech-
factor productivity of surgeons in an academic year and to evaluate the effect of surgical trainees on their productivity.
nical change, surgical training COMPETENCIES: Patient Care, Practice-Based Learning
and Improvement, Systems-Based Practice
STUDY DESIGN: We analyzed all the surgical procedures
performed from April 1 through September 30, 2013 in the Teikyo University Hospital. The nonradial and nonoriented Malmquist model under the variable returns-to-scale assumptions was employed. A decision-making unit is defined as a surgeon with the highest academic rank in the surgery. Inputs were defined as the number of physicians who assisted in surgery, and the time of surgical operation from skin incision to skin closure. The output was defined as the surgical fee for each surgery. April is the beginning month of a new academic year in Japan, and we divided the study period into April to June and July to September 2013. We computed each surgeon’s Malmquist index, efficiency change, and technical change. RESULTS: We analyzed 2789 surgical procedures that were performed by 105 surgeons. The Malmquist index of all surgeons was significantly greater than 1 (p ¼ 0.0033). The technical change was significantly greater than 1 (p o 0.0001). However, the efficiency change was not statistically significantly different from 1 (p ¼ 0.1817). CONCLUSIONS: The surgeons are less productive in the
beginning months of a new academic year. The main factor of this productivity loss is considered to be surgical training. C 2014 Association of Program ( J Surg 72:128-134. J Directors in Surgery. Published by Elsevier Inc. All rights reserved.)
This work was supported by JSPS KAKENHI Grant no. 25670337, which was given to Dr Yoshinori Nakata. Correspondence: Inquiries to Yoshinori Nakata, MD, MBA, Teikyo University Graduate School of Public Health, 2-11-1 Kaga, Itabashi-ku, Tokyo 173-8605, Japan; fax: (339) 63-2687; e-mail: [email protected]
128
INTRODUCTION Operating room efficiency is an important concern in most hospitals today.1 The most significant part of the operating room efficiency is considered to be surgical procedures because they usually occupy the longest time portion of the operating room time. When a new academic year starts, a number of new interns, residents, fellows, and attending physicians begin to work in a teaching hospital. It has been suggested that new trainees adversely affect the productivity in the operating rooms of a university hospital.2 In fact, it is demonstrated that surgical trainees’ participation in surgical procedures is associated with an increase in total operative time.3 Although those studies may be clinically significant, they only represent partial productivity measures, which evaluate only 1 factor (input) and 1 product (output), while assuming other variables as constant. A partial productivity measure may impute productivity changes to 1 input (or 1 output) that are really attributable to some other input (or output).4 However, there are no studies that have evaluated total factor productivity of surgeons that combines all inputs and all outputs. The Malmquist productivity index (Malmquist index, MI) represents total factor productivity change of a decision-making unit (DMU) between 2 periods under dynamic situation and is an example of in comparative statics analysis.4 It is based on data envelopment analysis (DEA), which evaluates relative efficiency of DMUs against the efficient frontier under static conditions in a single period. By comparing DEA results between 2 time periods, MI can divide productivity change into 2 components, one measuring efficiency change (EC) and the other measuring
Journal of Surgical Education & 2014 Association of Program Directors in Surgery. Published by 1931-7204/$30.00 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jsurg.2014.06.022
technical change (TC).5 MI is defined as the product of EC and TC terms. The EC term relates to the degree to which a DMU improves or worsens its efficiency, whereas the TC term reflects the change in the efficient frontiers between the 2 time periods. If productivity change of a DMU is compared between period 1 and period 2, they are mathematically defined as follows4: EC ¼ Efficiency of a DMU in period 2 with respect to period 2 frontier = Efficiency of a DMU in period 1 with respect to period 1 frontier TC ¼
Efficiency of a DMU in period 1 with respect to period 1 frontier = Efficiency of a DMU in period 1 with respect to period 2 frontier Efficiency of a DMU in period 2 with respect to period 1 frontier = Efficiency of a DMU in period 2 with respect to period 2 frontier 1=2
MI ¼ EC TC for example, we choose a 2-input, 1-output model that we will use in this study (Fig.). EC is defined as follows: EC ¼
OA=OB OE =OG
TC is defined as follows: OE=OG OC =OB 1=2 OE OC 1=2 ¼ TC ¼ OF =OG OA=OB OF OA thus, combining EC with TC, we get OC =OB OA=OB 1=2 MI ¼ EC TC ¼ OE=OG OF =OG
TABLE 1. Technical Terminology for Productivity Study Factor
The input used to produce goods and services Input Same as factor or resource Product The goods and services produced from the inputs Output Same as product or goods and services Total factor productivity Productivity that includes all the outputs produced and accounts for all of the inputs used to produce these outputs Decision-making unit The entity that is regarded as (DMU) responsible for converting inputs into outputs Data envelopment A frontier analysis method that analysis (DEA) evaluates relative efficiency of DMUs Efficient frontier A combination of minimum inputs that produces a certain amount of outputs, or a combination of maximum outputs that are produced from a certain amount of inputs Nonradial, nonoriented A variant of Malmquist model that Malmquist model takes account of all existing slacks (input excesses and output shortfalls) Variable returns to scale A production function that assumes that outputs increase variably when inputs increase proportionally Comparative statics An analysis that compares analysis 2 different economic outcomes, before and after a change
where OA, OB, OC, OE, OF, and OG are defined as the linear distances from the origin to the respective points. There have been a number of studies that applied DEA to surgery, and they provided clinically significant results.6-8 Although there has never been a study that applied a DEAbased MI model to surgery, it is considered appropriate to do so because MI is a nonparametric DEA model under time-dependent situations.4 Moreover, the MIs have been used to assess productivity change in a variety of sectors, such as agriculture, airlines, banking, electric utilities, insurance companies, and public sectors.5 Technical terminology used in efficiency and productivity study is summarized in Table 1. The goal of this study was to compute total factor productivity of surgeons in an academic year and to evaluate the effect of surgical trainees on their productivity change.
METHODS FIGURE. The Malmquist index in a 2-input, 1-output model. A DMU is in G in period 1 and moves to B in period 2. P1 indicates the efficient frontier in period 1. P2 indicates the efficient frontier in period 2.
The Teikyo University institutional review board approved our study. Anonymity of the data was strictly maintained by de-identification by the research team.
Journal of Surgical Education Volume 72/Number 1 January/February 2015
129
Data Teikyo University Hospital is located in the metropolitan Tokyo, Japan, serving a population of 1,000,000. It has 1152 beds and has a surgical volume of approximately 9000 cases annually. It has 13 surgical specialty departments. We collected data from surgical records in the Teikyo University Hospital electronic medical record system. One of the authors (Y.N.) looked up in person all the surgical procedures performed from April 1 through September 30, 2013 in the main operating rooms of Teikyo University Hospital and extracted the necessary information from the electronic medical record system. Exclusion criteria for surgery were as follows. First, surgical procedures performed under local anesthesia by surgeons were excluded to equalize resource utilization other than the inputs that we defined later. Oral, dermatologic, and ophthalmologic surgical procedures were excluded because most of their cases were minor surgeries performed under local anesthesia without anesthesiologists’ involvement, and those under general anesthesia do not represent the activity of their surgeons. Second, the surgical procedures were excluded if the patients die within a month after surgery to maintain the quality of surgery. Third, the surgical procedures that were not reimbursed under the current surgical payment system were excluded. Fourth, the surgical procedures were excluded if their records were incomplete for any reason. Analysis Framework We employed the nonradial and nonoriented Malmquist model under the variable returns-to-scale assumptions, which was particularly relevant because of its ability to employ multiple inputs and outputs simultaneously.9 In this analysis, we focused on the surgeons’ activity and their clinical decision. We defined in this study the DMU as a surgeon with the highest academic rank that scrubbed in the surgery. All the inputs and outputs are under the control of a DMU. Inputs were defined as (1) the number of physicians who assisted in surgery (assistants) and (2) the time of surgical operation from skin incision to skin closure (surgical time). The output was defined as the surgical fee for each surgery. It is classified as K000-K915 in the Japanese surgical fee schedule and is called “K codes.” Each surgical procedure is assigned to one of the K codes, which correspond with surgical fees. The fee is identical regardless of who (a senior surgeon or a surgical trainee) performs surgery as long as they have medical licensure, how many assistants they use, or how long it takes to complete surgery.10 The additional reimbursements for expensive surgical devices, such as automatic suture devices or imaging navigation devices, were excluded. Other fees for blood transfusion, medications, special insurance medical materials, and anesthesia were also excluded. The monetary values 130
of surgical fees were originally expressed in the Japanese yen, and were converted to US $ at $1 ¼ 100 yen to facilitate understanding by international readers. April is the beginning month of a new academic year in Japan. A number of new attending surgeons and trainees start working on April 1. They need some time to accustom themselves to a new operating room environment and new surgical instruments. In our hospital, new interns start working in May after a month’s orientation sessions. Therefore, we divided the study period into 2 periods: April to June 2013 (period 1) and July to September 2013 (period 2). We added all the inputs and outputs of the surgical procedures for each DMU during these study periods and computed his/her MI, EC, and TC using DEA-Solver-Pro Software (Saitech, Inc., Tokyo, Japan).4 MI 4 1 indicates progress in total factor productivity of the DMU from period 1 to 2, whereas MI ¼ 1 and MI o 1, respectively, indicate the status quo and deterioration in the total factor productivity. Similarly, a numerical value for the EC and the TC measure of greater than 1 implies that there is efficiency and technical progress, respectively. All the surgeons in the sample were given an MI, EC, and TC for each.11 All the surgeons analyzed were employees of Teikyo University and were salaried according to their ranks and experiences without any monetary incentives to increase surgical volume or productivity. The hospital charges surgeons’ surgical fees to Health Insurance Claims Review & Reimbursement Services, and the reimbursement becomes the revenue of the hospital. It pays surgeons their salary from this revenue. The surgeons belong to one of the following 10 surgical specialty departments: cardiovascular surgery, emergency surgery, general surgery, neurosurgery, obstetrics and gynecology, orthopedics, otorhinolaryngology, plastic surgery, thoracic surgery, and urology. We compiled their MIs, ECs, and TCs in their surgical specialties and calculated their means and standard deviations. We excluded from our analysis the surgeons who performed surgery in only 1 of these 2 periods. Statistical Analysis We used Excel Statistics 2008 Software (SSRI Co., Ltd., Tokyo, Japan) for our statistical analysis. Demographic data were analyzed within each study period with the 1-way analysis of variance. Multiple comparison tests were performed using the Bonferroni t tests.12 We also compared demographic data between both the study periods using the paired t tests. We compared the MIs, ECs, and TCs of all surgeons and those of each surgical specialty against 1 using the Student t test. The MIs, ECs, and TCs of each surgical specialty were compared against 1 using the Student t test. We also compared the MI, EC, and TC of each surgical specialty against other specialties using the 1-way analysis of variance.
Journal of Surgical Education Volume 72/Number 1 January/February 2015
TABLE 2. Demographic Data in Period 1 (April to June 2013) Specialty Cardiovascular surgery Emergency surgery General surgery Neurosurgery Obstetrics and gynecology Orthopedics Otorhinolaryngology Plastic surgery Thoracic surgery Urology All surgical procedures
DMUs 5 12 18 5 9 24 10 9 4 9 105
Cases 102 169 236 63 159 277 101 88 59 101 1355
Assistants Per Case 1.25 1.71 1.81 1.42 1.78 1.73 0.83 1.15 1.06 1.43 1.52
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
Time Per Case (min)
0.47 0.63 0.76 0.42 0.37 0.72 0.48║ 0.60 0.26 0.30 0.66
214 131 201 165 83 119 96 120 89 115 135
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
*
65 41 96‡ 68 35 46 41 58 25 60 70
Fee Per Case (US $) 6732 2544 3564 5775 2335 2660 1435 2100 5613 2082 3014
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
3697† 1364 1802 4194§ 977 1634 1313 1177 1847¶ 943 2200
Characteristics of each surgical specialty. Assistants per case, time pre case, and fee per case are expressed in mean ⫾ standard deviation. *The value is significantly different from obstetrics and gynecology and otorhinolaryngology (p o 0.05). † The value is significantly different from emergency surgery, general surgery, obstetrics and gynecology, orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). ‡ The value is significantly different from obstetrics and gynecology, orthopedics, otorhinolaryngology, thoracic surgery, and urology (p o 0.05). § The value is significantly different from obstetrics and gynecology, orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). The value is significantly different from emergency surgery, general surgery, obstetrics and gynecology, and orthopedics (p o 0.05). ¶ The value is significantly different from otorhinolaryngology (p o 0.05).
Multiple comparison tests were performed using the Bonferroni t tests. A p o 0.05 was considered statistically significant.12
RESULTS We analyzed 2789 surgical procedures performed by 105 surgeons (DMUs) during both the study periods. The demographic data of each surgical specialty during both the study periods were shown in Tables 2 and 3. There were significant differences among surgical specialties in assistants per case, surgical time per case, and surgical fee per case (p o 0.05). Between both the study periods, there were no significant differences in assistants per case, surgical
time per case, and surgical fee per case within each surgical specialty except assistants per case in emergency surgery (p ¼ 0.0012). The MIs, ECs, and TCs were shown in Table 4. The MI of all surgeons was significantly greater than 1 (p ¼ 0.0033), which demonstrated that the total factor productivity improved significantly from period 1 to period 2. The TC of all surgeons was significantly greater than 1 (p o 0.0001). However, the EC of all surgeons was not significantly different from 1 (p ¼ 0.1817). The natural logarithms of the MI, TC, and EC allow us to interpret these results as percent change.13 On average, the surgeons were 27.8% more productive in period 2 than in period 1. This productivity growth was decomposed into 3.0% increase in EC and 24.8% increase in TC.
TABLE 3. Demographic Data in Period 2 (July to September 2013) Specialty
Cases
Cardiovascular surgery Emergency surgery General surgery Neurosurgery Obstetrics and gynecology Orthopedics Otorhinolaryngology Plastic surgery Thoracic surgery Urology All surgical procedures
113 179 249 65 182 264 119 83 63 117 1434
Assistants Per Case 1.07 1.49 1.84 1.37 1.96 1.92 0.87 1.14 1.49 1.42 1.57
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
Time Per Case (min)
0.90 0.53† 0.61 0.46 0.24 0.73 0.50║ 0.59 0.24 0.68 0.69
186 121 186 180 93 104 96 122 107 119 129
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
92 50 67‡ 64 27 53 46 57 11 61 64
Fee Per Case (US $) 6125 3040 3665 6473 3182 2247 1427 1921 7259 1974 3108
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
4702* 2513 1414 3103§ 789 1709 1238 1425 563¶ 947 2377
Characteristics of each surgical specialty. Assistants per case, time per case, and fee per case are expressed in mean ⫾ standard deviation. *The value is significantly different from orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). † The value is significantly different from the corresponding value in period 1 (p ¼ 0.0012). ‡ The value is significantly different from obstetrics and gynecology, orthopedics, and otorhinolaryngology (p o 0.05). § The value is significantly different from emergency surgery, orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). The value is significantly different from general surgery, obstetrics and gynecology, and orthopedics (p o 0.05). ¶ The value is significantly different from emergency surgery, general surgery, obstetrics and gynecology, orthopedics, otorhinolaryngology, plastic surgery, and urology (p o 0.05). Journal of Surgical Education Volume 72/Number 1 January/February 2015
131
TABLE 4. Malmquist Index, Efficiency Change, and Technical Change of All Surgeons and of Each Surgical Specialty Specialty All surgeons Cardiovascular surgery Emergency surgery General surgery Neurosurgery Obstetrics and Gynecology Orthopedics Otorhinolaryngology Plastic surgery Thoracic surgery Urology
Malmquist Index 1.32 1.28 1.17 1.11 1.27 1.40 1.40 2.23 0.93 1.11 1.17
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
*
1.17 0.51 0.32* 0.34 0.39 0.78 1.65 2.49 0.38 0.24 0.58
Efficiency Change 1.03 1.22 0.98 0.97 1.12 1.01 1.09 1.81 0.76 1.03 0.89
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
0.87 0.41 0.24 0.31 0.19 0.23 1.06 2.14 0.42 0.11 0.24
Technical Change 1.28 1.05 1.20 1.16 1.12 1.45 1.45 1.33 1.30 1.07 1.28
⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾ ⫾
0.58* 0.18 0.17* 0.20* 0.20 0.89 1.01* 0.33* 0.25* 0.13 0.30*
*The value is significantly greater than 1 (p o 0.05).
The MI of emergency surgeons was significantly greater than 1 (p ¼ 0.043). The TCs of emergency (p ¼ 0.001), general (p ¼ 0.002), orthopedic (p ¼ 0.020), otorhinolaryngologic (p ¼ 0.005), plastic (p ¼ 0.003), and urologic surgeons (p ¼ 0.013) were significantly greater than 1. The difference in MIs, ECs, and TCs was not statistically significant among surgical specialties (p ¼ 0.5277, 0.4124, and 0.7959, respectively). Using the Bonferroni t tests, we did not find any significant difference in MIs, ECs, and TCs among surgical specialties, either.
DISCUSSION We demonstrated that the surgeons’ MI in the operating rooms was significantly greater than 1. This demonstrates that the surgeons’ productivity improved significantly from period 1 to period 2. This is the first study that evaluated the surgeons’ productivity using the Malmquist model. The surgeons were less productive in the beginning months of a new academic year than in the subsequent months. The 3-month study period was so short that any confounding factors were minimized. This suggests that the newly assigned trainees may be main factors of this productivity loss. The source of productivity growth is due to TC, not due to EC because TC was significantly greater than 1 and EC was not different from 1. This means that productivity growth is because surgical technique improved from period 1 to period 2.4 This finding is consistent with an initial progress of surgical technique of surgical trainees. However, their relative efficiency did not change from period 1 to period 2.4 Surgical training reduces surgeons’ productivity in 2 ways. First, it takes longer to complete surgery with participation of newly assigned trainees than those accustomed to their surgery.3 Second, the surgeons use more assistants as a means of education. The physicians who assist in surgery are usually surgical trainees in teaching hospitals. Those 2 variables were not statistically different between both the periods except the number of assistants in 132
emergency surgery when they were compared separately. However, the productivity demonstrated significant changes when the variables were evaluated simultaneously as demonstrated in total factor productivity measures. These 2 factors also suggest that the surgical training may negatively affect productivity of surgeons in the beginning of a new academic year. The current system for the reimbursement of surgical fees to hospitals in Japan is extremely vague, providing only total prices, adjusted to cover costs for each surgery with no explanation of component costs.14 In addition, the reimbursement system does not consider surgical training cost at all; the reimbursement is identical regardless of whether they are teaching hospitals or private hospitals.10 This reimbursement system discourages surgical training. Surgical training significantly reduces surgeons’ productivity, as demonstrated in this study. Therefore, the reimbursement should take surgical training cost into account and compensate the hospitals for the productivity loss of surgeons at least for 3 months in the beginning of a new academic year. We did not find any significant difference in MIs among surgical specialties. All the surgical specialties uniformly improved their productivity from the beginning months to the subsequent months. This suggests that the productivity loss of surgeons due to surgical training is common among all surgical specialties. Japanese hospitals should be compensated for the productivity loss of their surgeons of all surgical specialties. For example, the hospitals should be reimbursed with 27.8% more surgical fees for the first 3 months of an academic year to compensate for surgeons’ productivity loss because surgeons were 27.8% more productive on average in period 2 than in period 1 as our results demonstrated. As demonstrated in Tables 2 and 3, surgery is widely different in resource utilization and surgical fees among surgical specialties. These characteristics suggest that the Malmquist model that we employed was appropriate because of its ability to take simultaneously multiple different inputs and outputs into account.4 The simple comparison of surgical time, number of assistants, or surgical fees
Journal of Surgical Education Volume 72/Number 1 January/February 2015
does not reflect the heterogeneity of surgery among different surgical specialties. Its evaluation tends to be limited within a few similar types of common surgical procedures.3 Our method is better in aggregating heterogeneous surgical procedures and simultaneously comparing different surgical procedures by taking all inputs and outputs into account. We selected in this study only 2 inputs and a single output for surgery, although there are a number of other inputs and outputs of surgery. For example, anesthesia time is not included in our analysis because it is separately billed and reimbursed. Turnover time is not under surgeons’ control. We did not take into consideration technical difficulty of surgery, training period required to achieve clinical skills, or value of life for patients created by surgery in our analysis. For surgeons’ activity and their clinical decision, although surgical time and the number of assistants are objective measurements, technical difficulty and the value of life are not.14 The problem of technical difficulty regarding the patient’s condition requires an adjustment by patient or patient group, which makes interspecialty comparison impossible. However, the technical difficulty regarding the patient’s condition is not under the control of surgeons (DMUs). Moreover, the technical difficulty is expected to correlate positively with surgical time because the higher the technical difficulty, the longer the surgery. The training period required also positively correlates with technical difficulty. The correlated variables would only decrease the discriminatory power of the model without adding useful information.15 It is acceptable to eliminate one of the factors that are correlated with each other.16 Therefore, we selected these variables, although we know that they do not reflect all aspects of surgery. The effect of more surgical assistants is not necessarily in the operating room, but in other parts of the hospital. If they are tied up in the operating room, they cannot do any other work in the ward or outpatient clinic. If the third and fourth assistants do not help surgeons, they should instead work in the other part of the hospital. Therefore, we assumed from the hospital’s viewpoint that the fewer the assistants, the more productive the surgeons were. There are some limitations in our study. First, this is a study conducted in a single large teaching hospital in Tokyo, Japan. Our surgeons may not represent all the surgeons in Japan. However, this is the first study to evaluate the surgeons’ total factor productivity using the actual data in the Malmquist model. Second, this study is based on the Japanese surgical reimbursement system. Our results may be invalid under a reimbursement system that is different from the Japanese system. Third, we compared the first 3 months of a new academic year to the subsequent 3 months; therefore, the productivity improvement may be due to other factors, such as seasonal case-mix variation. However, as seen in the demographic data, there were no significant differences between both the study periods in inputs and outputs within each surgical specialty, except
assistants per case in emergency surgery. Moreover, each surgeon performs only 5 to 10 types of surgical procedures at most, and his/her operative activity is fairly constant because we studied 2 consecutive 3-month periods. It is expected that there is no significant difference in case-mix variation. Furthermore, 3-month periods are short enough to minimize any other confounding factors, such as change in surgeons, surgical practice patterns, nursing practice, and reimbursement system. Therefore, 3-month study periods are appropriate and the main factor of this productivity change is speculated to be due to surgical training. Fourth, we simply considered the number of assistants without taking their experience into account. It is obvious that a surgeon of lesser rank or a newly appointed senior resident is not equivalent to junior surgical trainees in assisting surgery. An unequal distribution of residents with different experience is also likely. On services like cardiovascular, thoracic, and neurosurgery, more senior trainees need to be present, and it is unlikely that a junior trainee would get to do anything technically, or perhaps even scrub. However, the detailed data were unavailable.
CONCLUSION We demonstrated by the Malmquist model that the surgeons are less productive in the beginning months of a new academic year. The main factor of this productivity loss may be due to surgical training.
REFERENCES 1. Eappen S, Flanagan H, Bhattacharyya N. Introduction
of anesthesia resident trainees to the operating room does not lead to changes in anesthesia-controlled times for efficiency measures. Anesthesiology. 2004;101(5): 1210-1214. 2. Posner KL, Freund PR. Resident training level and
quality of anesthetic care in a university hospital. Anesth Analg. 2004;98(2):437-442. 3. Papandria D, Rhee D, Ortega G, et al. Assessing
trainee impact on operative time for common general surgical procedures in ACS-NSQIP. J Surg Educ. 2012;69(2):149-155. 4. Cooper WW, Seiford LM, Tone K. Data Envelop-
ment Analysis: A Comprehensive Text With Models, Applications, References and DEA-Solver Software. 2nd ed New York: Springer, 2007. 328-347. 5. Coelli TJ, Rao DSP, O’Donnell CJ, Battese GE. An
Introduction to Efficiency and Productivity Analysis. 2nd ed New York: Springer, 2005. 289-310, 76-77.
Journal of Surgical Education Volume 72/Number 1 January/February 2015
133
6. Dexter F, O’Neill L. Data envelopment analysis to
11. Hollingsworth B, Peacock SJ. Efficiency Measurement in
determine by how much hospitals can increase elective inpatient surgical workload for each specialty. Anesth Analg. 2004;99(5):1492-1500.
Health and Health Care. Oxon: Routledge, 2008. 37-39.
7. Basson MD, Butler T. Evaluation of operating room
suite efficiency in the Veterans Health Administration system by using data-envelopment analysis. Am J Surg. 2006;192(2):649-656. 8. O’Neill L, Dexter F. Tactical increase in operating
room block time based on financial data and market growth from data envelopment analysis. Anesth Analg. 2007;104(2):355-368. 9. Thanassoulis E, Portela MCS, Despic O. Data
envelopment analysis: the mathematical programming approach to efficiency analysis. In: Fried HO, Lovell CAK, Schmidt SS, eds. The Measurement of Productive Efficiency and Productivity Growth. New York: Oxford University Press, 2008. 251-420. 10. Social Insurance Institute (Shakaihoken Kenkyuusho).
Interpretation of medical reimbursement scales, April 2012 (Ikatensuuhyou no Kaishaku). 2012: 622-780 [in Japanese].
134
12. Glantz SA, Slinker BK. Primer of Applied Regression
and Analysis of Variance. New York: McGraw-Hill; 1990. 272-313. 13. Fare R, Grosskopf S, Lindgren B, Roos P. Productivity
developments in Swedish hospitals: a Malmquist output index approach. In: Charnes A, Cooper W, Lewin AY, Seiford LM, eds. Data Envelopment Analysis: Theory, Methodology and Applications. New York: Springer, 1994. 253-272. 14. Hayashida K, Imanaka Y. Inequity in the price of
physician activity across surgical procedures. Health Policy. 2005;74:24-38. 15. Paradi JC, Yang Z, Zhu H. Assessing bank and bank
branch performance; modeling considerations and approaches. In: Cooper WW, Seiford LM, Zhu J, eds. Handbook on Data Envelopment Analysis. 2nd ed. New York: Springer, 2011. 315-362. 16. Ozcan YA. Health Care Benchmarking and Perform-
ance Evaluation: An Assessment Using Data Envelopment Analysis. New York: Springer; 2008.
Journal of Surgical Education Volume 72/Number 1 January/February 2015
