Total Joint Arthroplasty Cost Savings With a Rapid Recovery Protocol in a Military Medical Center

MILITARY MEDICINE, 177, 1:64, 2012 Total Joint Arthroplasty Cost Savings With a Rapid Recovery Protocol in a Military Medical Center MAJ David M. Doman, MC USA ; COL Tad L. Gerlinger, MC USA ABSTRACT The reported short-term benefits of rapid recovery protocols for total joint arthroplasty primarily come from specialized centers of excellence. The feasibility of achieving similar benefits at a military health care facility has not been reported. The authors hypothesized that application of such a protocol in this setting would decrease hospital stay and costs. A retrospective study was conducted comparing 85 hip and knee replacements by one surgeon using conventional protocol to 90 cases by a second surgeon using a rapid recovery protocol in the same hospital. Outcome measures included operative time, length of hospital stay, pain at discharge, use of inpatient rehabilitation facilities, complications requiring readmission, and inpatient admission costs. The results showed decreased length of stay by 2.9 days ( p < 0.001) in the rapid recovery group, resulting in average cost savings of $1,511 ( p < 0.001) with shorter operative time, equivalent pain at discharge, and fewer discharges to rehabilitation facilities. This feasibility study shows promising results, but prospective randomized trials are necessary to draw firm conclusions on the superiority of a rapid recovery protocol for total hip and knee arthroplasty in a military medical system. INTRODUCTION Over the last several years, many efforts have been made to improve outcomes in total knee and hip arthroplasty, two procedures which already demonstrated reproducibly excellent results. One target area of improvement has been accelerating patients recovery from surgery. 1 6 Some authors focus on smaller incision size for hip and knee arthroplasty and the presumed decreased trauma to soft tissue. 3,7 A broader approach is described by others, who endorse a multimodal clinical pathway incorporating preoperative patient education to manage patient expectations, aggressive preoperative and intraoperative pain management, minimization of surgical exposure, and early postoperative mobilization. 8 11 Advocates of this and other similar rapid recovery protocols claim less intraoperative blood loss, improved earlier pain control, earlier hospital discharge, decreased use of rehabilitation facilities, less dependence on narcotics, less use of assistive devices, and improved patient satisfaction without compromising functional outcome.6,8,12 16 These efforts have a similar goal of speeding the patient s recovery from surgery and return to function. In addition to benefiting the patient, sooner discharge to home and less reliance of inpatient rehabilitation facilities represent a significant cost savings to the health care system. The majority of authors citing improvement in these key outcomes practice in hospitals with a special commitment toward total joint arthroplasty. 6,8,12 16 In such settings, dedicated resources are often available to optimize the workflow, Department of Orthopedics & Rehabilitation, Brooke Army Medical Center, 3851 Roger Brooke Drive, Fort Sam Houston, TX 78234. Portions of this material were presented as podium presentations on three separate occasions: Annual Meeting of Society of Military Orthopedic Surgeons in Denver, CO, in December, 2007; University of Texas Health Science Center Roy Davis Competition in San Antonio, TX, in May, 2011; and San Antonio Military Medical Center Omer Day in San Antonio, TX, in May, 2011. resulting in streamlined patient care and decreased cost to the institution. These resources include specialized operating room support staff, anesthesia providers, nurses, and physical therapists. Presumably, these additional resources create ideal conditions to allow the cited benefits to be realized. However, the applicability of a rapid recovery protocol at a military health care facility is unclear. The resources available to the arthroplasty surgeon at such a facility are significantly diminished when compared to the before mentioned specialized hospitals. This study will compare conventional techniques to a rapid recovery protocol for arthroplasty in the context of a military health care setting to see if similar outcomes and cost savings can be achieved. MATERIALS AND METHODS Data Collection The Institutional Review Board at the sponsoring institution approved this study. A retrospective chart review was conducted comparing 85 consecutive cases for a fellowship-trained arthroplasty surgeon using conventional surgical techniques and perioperative care (group 1) to 90 consecutive cases by a fellowship-trained arthroplasty surgeon (senior author) using a rapid recovery protocol (group 2) in the same managed care hospital. Inclusion criteria were primary elective total knee or hip arthroplasty performed at Brooke Army Medical Center during the study period. Two equal 12-month time periods served as data collection periods for each surgeon from April 2004 to March 2005 for the conventional technique and from April 2006 to March 2007 for the rapid recovery protocol group. These time periods were selected on the basis of finding two equivalent time periods where both surgeons were not deployed overseas and had a full caseload. All cases that fit the inclusion criteria were included in the study. Data gathered from the electronic inpatient database included duration 64 MILITARY MEDICINE, Vol. 177, January 2012

of surgery, length of hospital stay in days, pain level at time of discharge (Visual Analog Scale), discharge location (home with outpatient physical therapy versus to an inpatient rehabilitation facility), and complications requiring readmission within 30 days of surgery. Additional data collected included age, gender, race, body mass index (BMI), diagnosis, joint replaced (hip or knee), American Society of Anesthesiologists (ASA) physical status classification, day of week the operation took place, and type of anesthesia used. Cost of inpatient stay was determined using the Medical Expense Performance Reporting System (MEPRS), the standard cost estimation protocol for the U.S. military health care system. Clinical Protocols The conventional group, group 1, serves as a historical control. The group 1 surgeon used a standard posterior approach to the hip and subvastus approach to the knee. No preoperative analgesia was used. Selection of anesthesia modality was determined by the patient and anesthesiologist and not set by surgeon protocol. Surgical technique did not emphasize smaller incisions. There was no established postoperative pain regimen. Patients were discharged once pain was controlled on oral medications. Selected patients were discharged to inpatient rehabilitation facilities based on surgeon and therapist assessment of rehabilitation needs. Patients in group 2 followed a rapid recovery protocol starting with the preoperative period, where patient education in the form of clinical counseling and written material helped guide patient expectations toward a quick return to function after surgery, including ambulation on the first postoperative day. Preoperative pain medication was initiated the morning of surgery with celecoxib and oxycodone. Epidural anesthesia with intravenous sedation was encouraged and use of regional blocks was discouraged, but the ultimate choice for anesthesia type was determined by the anesthesia provider in consultation with the patient. The surgery was performed via a mini-posterior approach for hips and a mini-medial parapatellar approach for knees, unless individual patient factors dictated otherwise. The surgical incision length was not preestablished, but efforts were made to use the minimal length required to achieve optimal component position. The operating surgeon has an average incision length of less than 10 cm. Postoperatively, patients were continued on oxycodone and started on either hydrocodone/acetaminophen or oxycodone/ acetaminophen orally with intravenous morphine for breakthrough pain. Patient-controlled analgesia intravenous pumps were avoided. Foley catheter and epidural were removed the morning following surgery. When used, surgical drains were also removed postoperative day 1 unless excessive drainage was noted. The goal of ambulation on the first postoperative day was established with the patient preoperatively and inpatient physical therapists were utilized. The discharge criteria and decision to send patients to an inpatient rehabilitation facility was the same for both groups. Statistical Analysis A power analysis was completed before the study, with the assistance of SPSS Sample Power 2.0. According to this method, 64 subjects per group were required to detect a moderate (0.5 SD) effect size with a power of 80% and a level of confidence of 95%. This effect size, when applied to the dependant variables, is equivalent to 14 minutes of operative time, 0.6 days as an inpatient, a pain level difference of 1 point on a 10-point scale, and $916.80 in cost savings. With 85 and 90 subjects in the two groups, respectively, this study exceeds the power analysis requirements, with numbers sufficient to detect a 0.42 SD effect size. The independent variable is method of treatment for total joint replacement (conventional versus rapid recovery). The dependent variables are duration of surgery, length of hospital stay, pain level at time of discharge, discharge to home versus inpatient rehabilitation facility, number of complications requiring readmission within 30 days of surgery, and cost of inpatient stay. To test the hypothesis that there is difference between treatment arms in patient age, BMI, ASA classification, duration of surgery, length of hospital stay, postoperative pain, or cost of inpatient stay, two-tailed independent sample t-tests were performed. The ASA and cost variables were not normally distributed, so the results were confirmed with the Mann Whitney rank sum test. The length of stay variable had unequal variances, so the results were also confirmed with a Mann Whitney rank sum test. To test the hypothesis that there is a difference between treatment arms regarding patient gender, ethnicity, diagnosis, joint undergoing arthroplasty, anesthesia type, day of week of surgery, and discharge status, c 2 contingency tests were performed. Because some anesthesia and ethnicity categories had an expected count less than 5, the results were confirmed with a Fisher s exact test. To assess if length of stay correlated with day of the week, independent of treatment, a one-way analysis of variance test was performed. To assess if the change of length of stay between treatments could be ascribed to treatment or anesthesia alone, a two-factor analysis of variance test was performed. This was done with five categories of anesthesia initially based on the collected data, then with two categories of anesthesia based on whether or not the patient received general anesthesia. To test the hypothesis that the discharge destination is correlated with ASA, age, and BMI independent of treatments, Mann Whitney and independent sample t-tests were performed. There were three patients with complications requiring readmission, two for group 1 and one for group 2, so these results are simply reported and not subjected to statistical analysis because of the low numbers. RESULTS Data on patient characteristics were collected to ensure the two groups being compared were similar. Statistical analysis demonstrated that the two groups were similar in age, MILITARY MEDICINE, Vol. 177, January 2012 65

TABLE I. Patient Characteristics by Treatment Arm Group 1 ( n = 85) Group 2 ( n = 90) Type I Error Mean Age (Years) 61.5 58.9 p = 0.152 Mean ASA Class 2.3 2.3 p = 0.717 Mean BMI (kg/m 2 ) 29.7 31.6p = 0.026 Gender (Percentage Male Percentage Female) 42.3 57.6 62.2 37.8 p = 0.013 Region (Percentage Hips Percentage Knees) 44.7% 55.3% 38.9% 1.1% p = 0.531 Ethnicity (Caucasian African Asian Other) 67 11 0 7 66 12 1 11 p = 0.496 Diagnosis (Osteoarthritis Avascular Necrosis 75 5 2 2 1 78 5 4 2 1 p = 0.971 Dysplasia Inflammatory Arthritis Post-Traumatic) Day of Week of Operation (Mon-Tue-Wed-Thu-Fri) 1 45 6 30 3 5 52 2 30 1 p = 0.213 TABLE II. Type of Anesthesia by Treatment Arm Five Categories Group 1 ( n = 85) Group 2 ( n = 90) General Anesthesia 31 (36.5%) 15 (16.7%) Neuroaxial (Epidural or Spinal ) 21 (24.7%) 51 (56.7%) Anesthesia General Anesthesia and 20 (23.5%) 0 Regional Block Regional Block 10 (11.8%) 0 General and Neuroaxial Anesthesia 3 (3.5%) 24 (26.7%) Two Categories Group 1 ( n = 85) Group 2 ( n = 90) Received General Anesthesia 54 (63.5%) 39 (43.3%) Did Not Receive General 31 (36.5%) 51 (56.7%) Anesthesia Total 85 (100%) 90 (100%) ethnicity, joint replaced, ASA class, and diagnosis. Group 2 has a larger number of male patients and a higher average BMI. Table I shows these results. Anesthesia used was also analyzed, and there was a significant divergence in the methods used between the two groups. This was not unexpected based on the clinical protocols. The anesthesia data was grouped under five different categories and then simplified into two categories based on the use of general anesthesia ( Table II ), as this was hypothesized to be the most significant factor that might affect the dependant variables. Group 1 had a higher percentage of general anesthesia, and group 2 used more neuroaxial anesthesia. No regional blocks were used on group 2 patients. The outcome measures analysis showed that the patients in group 2 spent on average 20.3 minutes less time in the operating room and were discharged from the hospital 2.9 days earlier and with no statistical difference in pain level when compared to group 1. Group 2 patients were also less likely to require discharge to an inpatient rehabilitation facility. Two patients from group 1 were readmitted within 30 days, one for a typical chest pain and a second for revision surgery. One patient from group 2 was readmitted for intractable vomiting. Table III summarizes this data. Further analysis showed that difference in the type of anesthesia used between the two treatments could not account for the significant difference in length of stay. The difference in mean length of stay between groups 1 and 2 was 2.9 days. Figure 1 plots length of stay versus the five different categories of anesthesia for both treatment groups. Only three categories could be directly compared as no regional blocks were used in group 2. The two treatment groups, when matched for neuroaxial anesthesia, general anesthesia, and combined anesthesia, have a length of stay difference of 2.9, 2.6, and 3.5 days, respectively. The anesthesia variable was subsequently reorganized into general anesthesia and no general anesthesia, resulting in more equivalent groups for comparison. The two treatment groups, when matched for general anesthesia and no general anesthesia, both have a length of stay difference 2.8 days. ( Fig. 2 ) There is no significant interaction between treatment group and anesthesia for either the two or five category analysis ( p = 0.585, p = 0.979), which means the trends are parallel. There was no significant difference in day of the week the two surgeons operated ( p = 0.213), therefore, it is an unlikely cause for the difference in length of stay or difference in the rate of discharge to inpatient rehabilitation facilities (Table I ). Since discharge to inpatient rehabiliation facilities was less likely in group 2 (see above), other variables that might be associated with use of inpatient rehabilitation were assessed, revealing that increased age and increased ASA classification were associated with inpatient rehabilitation use ( p < 0.001, p = 0.007) and increased BMI was not ( p = 0.868) (Table IV ). However, age ( p = 0.152) and average ASA classification ( p = 0.717) were not statistically different between the two groups ( Table I ). Cost of inpatient stay was divided into seven categories according to the MEPRS protocol. Direct costs are those associated with providing direct medical care (physician and nursing care are examples), whereas indirect costs are those associated with other necessary but nonmedical expenses (housekeeping and utilities are examples). Surgical, ICU, radiology, and laboratory costs were also estimated. Ancillary costs represent other aspects of care to include pharmacy and pathology. The results are listed in Table V. DISCUSSION An assessment of the study population shows the two treatment groups to be similar in most variables measured. Two 66 MILITARY MEDICINE, Vol. 177, January 2012

TABLE III. Outcome Measures Group 1 ( n = 85) Group 2 ( n = 90) Difference 95% Confidence Intervals Type I Error Mean Surgery Duration 173.1 minutes 152.8 minutes 20.3 minutes 12.5 28.0 minutes p < 0.001 Mean Length of Stay 5.5 days 2.6 days 2.9 days 2.4 3.1 days p < 0.001 Pain Level at Discharge 2.2/10 2.6/10 0.4/10 0.3 1.0 p = 0.27 Discharged to Inpatient Rehab 28 patients (32.9%) 11 patients (12.2%) 20.7% N/A p = 0.002 Readmission Within 30 Days 2 patients (2.4%) 1 patients (1.1%) 1.3% N/A N/A FIGURE 1. Length of hospital stay versus anesthesia (five categories) by treatment group. FIGURE 2. Length of hospital stay versus anesthesia (two categories) by treatment group. exceptions include BMI and gender, with group 2 having an increased number of males and a more obese population. These differences in the two groups are difficult to explain as both surgeons drew from the same population, and the effect on the study s results is unclear. A significant finding is the decrease in hospital stay in group 2, as it is an objective outcome that represents a significant cost savings. An average decrease of 2.9 days over 90 patients represents approximately 260 patient-days over the course of a year, and at a mean cost savings of $1511 per patient, a cost savings of $135,990 a year. The validity of this outcome measure is supported by equivalent pain level at discharge (countering a potential criticism that group 2 patients were discharged with a greater degree of pain) and a decreased discharge to inpatient rehabilitation (countering an argument that the decreased length of stay in group 2 could be attributed to sending patients preferentially to inpatient rehabilitation). Another potential argument is that cost savings in earlier hospital discharge are offset by increased operative time and an increased rate of readmission. The observed decreased operative time for group 2 counters this argument, and readmissions were rare. Another potential criticism in the study is that length of stay may be dependent on which day of the week the operation was performed. Because natural disincentives exist to discharge over the weekend, this might tend to increase the hospital stay of patients who underwent surgeries later in the week. An analysis of data shows such an effect is unlikely. In the population studied, there was no significant difference between groups regarding day of operation ( p = 0.213) and, therefore, no difference in length of stay between groups can be attributed to difference in day of surgery ( Table I ). Another possible flaw in the study is the significant difference in anesthesia modalities used by the two groups. One might argue that the effect observed in the study is because of the type of anesthesia used. We tested this hypothesis by adjusting the data for the difference in anesthesia to determine if anesthesia modality alone could account for the difference between the two groups in length of hospital stay. When the two treatment arms were corrected for the differences in anesthesia, grouped both under the initial five anesthesia categories or simplified into two categories ( Figs.1 and 2 ), there remained a significant difference in hospital stay. This supports the hypothesis that anesthesia type alone could not account for the difference in hospital stay between groups. Because group 2 did have a lower rate of discharge to inpatient rehabilitation facility, the authors sought to find other variables that could be shown to influence use of inpatient rehabilitation. It was hypothesized that increasing obesity, age, and comorbidities would also predispose patients toward use of such facilities. Analysis of the data did not show a correlation with BMI and discharge to inpatient rehabilitation facilities, but did show such an association with increasing age and ASA classification ( Table III ). Since the ages and ASA for the two groups populations were equivalent, this supports the argument that the difference observed is because of the treatment protocol used. There are several limitations to the design of this study. The two treatment protocols were performed by two different MILITARY MEDICINE, Vol. 177, January 2012 67

TABLE IV. Discharge Location Compared to Selected Variables Variable Description n Home Inpatient Rehab Type I Error BMI Mean (kg/m 2 ) 175 30.7 30.5p = 0.868 Age Mean (Years) 175 57.5 69.5 p > 0.001 ASA Classification I 10 10 (7.4%) 0 p = 0.007 II 104 85 (62.5%) 19 (48.7%) III 61 41 (30.1%) 20 (51.3%) Treatment Arm Group 1 85 57 (67.1%) 28 (32.9%) p = 0.002 Group 2 90 79 (87.8%) 11 (12.2%) TABLE V. Inpatient Costs Assessment Group 1 ( n = 85) Mean Cost Group 2 ( n = 90) Mean Cost Difference Type I Error Direct Costs $1.735.83 $1059.40 $676.43 p < 0.001 ICU Costs $87.76 $83.13 $4.63 p > 0.05 Support Costs $2.760.39 $1739.09 $1021.30 p < 0.001 Surgical Costs $6742.83 $6125.12 $617.71 p < 0.001 Radiology Costs $31.66 $38.32 $6.66 p < 0.001 Laboratory Costs $164.76 $152.22 $12.54 p < 0.001 Ancillary Costs $2917.48 $3731.83 $814.35 p < 0.001 Total Costs $14440.71 $12929.11 $1511.60 p < 0.001 surgeons, which introduces a potential performance bias. The two groups were not assessed concurrently, but over a time period spanning 3 years, again adding performance bias as new techniques are developed over time. Hip and knee arthroplasty are compared together in this study, and although the ratio is not significantly different between groups, this introduces selection bias. The conventional technique is not strictly defined when compared to the rapid recovery protocol. The incision lengths are not compared in this study because of the retrospective nature of the study and the broader focus on rapid recovery protocol. Early complications assessment is admittedly limited with visibility only of complications requiring readmission. There is also no measure of mid- to long-term outcomes of the two surgical protocols such as infection, dislocation, or need for revision. The most important factor in the survivability and success of total joint arthroplasty is correct position of components. This study could be expanded in the future to assess postoperative radiographs and mid- to long-term outcomes. This would provide further information regarding the effect of the surgical technique alone, although the performance bias would remain. This study focused on short-term outcomes only. At our institution, we have adopted a broad rapid recovery protocol to improve recovery time from total joint arthroplasty. We do not claim that these results are solely because of the incision size, nor do we attempt to separate the effects of each factor in the rapid recovery protocol, other than the anesthesia modality discussion above. Further studies may help differentiate the importance of each effect. This study was preformed primarily as a feasibility study focusing on joint arthroplasty in a military health care environment. Despite these shortcomings, the stated goals of the study were met. With the application of a rapid recovery protocol for total joint arthroplasty in a managed care environment, a significant decrease in length of hospital stay and therefore cost savings was achieved. This feasibility study shows promising results, but prospective randomized trials are necessary to draw firm conclusions on the superiority of a rapid recovery protocol for total hip and knee arthroplasty in a military medical system. ACKNOWLEDGMENT The authors appreciate the contribution of John Ward, PhD, for assistance with statistical analysis. REFERENCES 1. Berger, RA : Mini-incisions: two for the price of one! Orthopedics 2002 ; 25 (5) : 472, 498. 2. Berry DJ, Berger RA, Callaghan JJ, et al : Minimally invasive total hip arthroplasty. Development, early results, and a critical analysis. J Bone Joint Surg Am 2003 ; 85: 2235 46. 3. Goldstein WM, Branson JJ, Berland KA, Gordon AC : Minimalincision total hip arthroplasty. J Bone Joint Surg Am 2003 ; 85 (Suppl 4) : 33 8. 4. Hungerford DS : Minimally invasive total hip arthroplasty: in opposition. J Arthroplasty 2004 ; 19 (4 Suppl 1) : 81 2. 5. Sculco TP : Minimally invasive total hip arthroplasty: in the affirmative. J Arthroplasty 2004 ; 19 (4 Suppl 1) : 78 80. 6. Wenz JF, Gurkan I, Jibodh SR : Mini-incision total hip arthroplasty: a comparative assessment of perioperative outcomes. Orthopedics 2002 ; 25 (10) : 1031 43. 7. Wright JM, Crockett HC, Delgado S, Lyman S, Madsen M, Sculco TP : Mini-incision for total hip arthroplasty: a prospective, controlled investigation with 5-year follow-up evaluation. J Arthroplasty 2004 ; 19 (5) : 538 45. 8. Berger RA, Jacobs JJ, Meneghini RM, Della Valle C, Paprosky W, Rosenberg AG : Rapid rehabilitation and recovery with minimally invasive total hip arthroplasty. Clin Orthop Relat Res 2004 ; 429: 239 47. 68 MILITARY MEDICINE, Vol. 177, January 2012

9. Inaba Y, Dorr LD, Wan Z, Sirianni L, Boutary M : Operative and patient care techniques for posterior mini-incision total hip arthroplasty. Clin Orthop Relat Res 2005 ; 441: 104 14. 10. Pagnano MW, Hebl J, Horlocker T : Assuring a painless total hip arthroplasty: a multimodal approach emphasizing peripheral nerve blocks. J Arthroplasty 2006 ; 21 (4 Suppl 1) : 80 4. 11. Peters CL, Shirley B, Erickson J : The effect of a new multimodal perioperative anesthetic regimen on postoperative pain, side effects, rehabilitation, and length of hospital stay after total joint arthroplasty. J Arthroplasty 2006 ; 21 (6 Suppl 2) : 132 8. 12. Chimento GF, Pavone V, Sharrock N, Kahn B, Cahill J, Sculco TP : Minimally invasive total hip arthroplasty: a prospective randomized study. J Arthroplasty 2005 ; 20 (2) : 139 44. 13. Dorr LD, Maheshwari AV, Long WT, Wan Z, Sirianni LE : Early pain relief and function after posterior minimally invasive and conventional total hip arthroplasty. A prospective, randomized, blinded study. J Bone Joint Surg Am 2007 ; 89 (6) : 1153 60. 14. Howell JR, Masri BA, Duncan CP : Minimally invasive versus standard incision anterolateral hip replacement: a comparative study. Orthop Clin North Am 2004 ; 35 (2) : 153 62. 15. Schroer WC, Diesfeld PJ, Reedy ME, Lemarr AR : Mini-subvastus approach for total knee arthroplasty. J Arthroplasty 2008 ; 23 (1) : 19 25. 16. Sculco TP, Jordan LC, Walter WL : Minimally invasive total hip arthroplasty: the hospital for special surgery experience. Orthop Clin North Am 2004 ; 35 (2) : 137 42. MILITARY MEDICINE, Vol. 177, January 2012 69