Background
Enhanced recovery after surgery (ERAS) programs, also known as enhanced recovery programs (ERPs), fast track, or rapid recovery, are based on the work of Henrik Kehlet in colorectal surgery (Kehlet, 1997). In recent years, there has been considerable interest in their introduction in orthopaedic surgery, especially in hip and knee replacements (Ibrahim, Twaij, Giebaly, Nizam, & Haddad, 2013; Kehlet & Thienpont, 2013). Programs are designed to prepare patients for, and reduce the total impact of, surgery, helping them recover more quickly. These programs include preoperative information and optimization of comorbidities, anesthetic and postoperative analgesia, surgical technique, perioperative blood management, early mobilization, rehabilitation, and discharge planning (Kehlet & Thienpoint, 2013). Such programs take a whole-system, evidence-based, multidisciplinary approach (Ibrahim et al., 2013; Kehlet & Thienpoint, 2013). In orthopaedics, they may reduce mortality, average length of stay (ALOS), and perioperative complications without an increase in complication or readmission rates (Kearney, Jennrich, Lyons, Robinson, & Berger, 2011; Malviya et al., 2011; McDonald, Siegmeth, Deakin, Kinninmonth, & Scott, 2012; Wainwright & Middleton, 2010). As there are approximately six times more elective hip and knee replacements performed per year in the United Kingdom than colorectal procedures, the potential benefits may be greater (Wainwright & Middleton, 2010).
Despite the success of ERAS programs, there have been few published results in Australasia (Keane et al., 2012). We developed and implemented ERAS protocols as part of a wider program-the Orthopaedic Patient Pathway funded by the Ministry of Health, under the Elective Services Productivity and Workforce Program. The purpose of this article is to describe the development, implementation, and results of the first 18 months of the program in our public hospital. The first 528 patients undergoing primary hip or knee replacement following the introduction of the ERAS program were compared with a historical cohort from the 18-month period immediately prior to their introduction.
Methods
Our institution is the main hospital for a population of 200,000, covering a sparsely populated area of 32,000 square kilometers. There are 10 orthopaedic surgeons and one arthroplasty fellow who perform approximately 400 hip and knee arthroplasties including revision surgery per year. Elective patients are admitted to a single orthopaedic elective ward (20-24 beds). In addition to the normal ward nursing staff, there is one permanent orthopaedic physiotherapist, a rotating physiotherapist, and one occupational therapist. Key members of the department (including the elective ward charge nurse, the senior permanent physiotherapist, and the clinical leader) attended a 1-day workshop on the principles and implementation of enhanced recovery (Wainwright & Middleton, 2010). Having been enthused by the presentation, the team members resolved to introduce the techniques to our institution. Over the next 6 months, all aspects of the patient journey were reviewed. Small-scale audits were conducted on cancellations at the preadmission clinic and on day of surgery, effect of preoperative anemia on transfusion requirements and ALOS, use of drains, use of femoral nerve blocks, and local anesthetic infiltration techniques. New protocols were developed by the nursing, physiotherapy, and medical staff. An experienced former orthopaedic charge nurse was appointed to a part-time ERAS facilitator role, and the head orthopaedic surgeon was the clinical champion for the project.
An audit of cancellations at the preadmission clinic showed the commonest reasons for cancellation of surgery were poorly controlled medical conditions, need for dental care, and skin problems such as ulcers. This led to the development of a preoperative health questionnaire concentrating specifically on recent dental care, skin lesions, and chronic health conditions. This was mailed to patients and general practitioners (GPs) prior to the preadmission appointment. Problems were identified and addressed, and surgery was delayed if necessary. An audit of preoperative anemia showed that patients who were anemic (males <130 g/L, females <120 g/L) stayed a mean 1.6 days longer than those patients who were not anemic (6.8 vs. 5.2 days) and 56% of anemic patients required transfusion compared with 12.5% of nonanemic patients. This led to the development, in conjunction with the Hematology Department, of an algorithm for the early identification and management of preoperative anemia. Facilities and arrangements for intravenous iron infusion and oral iron supplementation were developed for patients with preoperative iron-deficient anemia.
Community service physiotherapists and occupational therapists had seen patients from the historical cohort for education and aids such as raised toilet seats and crutches. Although education was given, this was not always consistent or up to date. A new preoperative education class was developed for all patients who lived within 1-hour travel time from our hospital and run by members of the elective ward staff. Aids were issued, and patients were instructed on preoperative physiotherapy exercises. Details of their postoperative care and expectations were given by the staff who would be looking after them, which ensured continuity. The patient information guide was rewritten to reflect a shifting of responsibility onto the patient and the family in areas such as smoking cessation and discharge planning.
Standardized anesthetic and analgesia guidelines were developed with a consultant anesthetist (see Table 1). The goal was for mobilization of the patient on the day of surgery if possible and better management of postoperative pain. Key initiatives included the routine administration of tranexamic acid (if no contraindications) and increased use of parenteral and oral nonsteroidal anti-inflammatory drugs. A randomized study we conducted showed no advantages to postoperative continuous femoral nerve infusions over a single-shot femoral nerve block, so their use was discouraged (Wyatt, Wright, Locker, Stout, & Theis, 2015). In turn, femoral nerve block has been superseded by local infiltration with ropivacaine, adrenaline, and tranexamic acid in patients undergoing knee replacements.
Discharge criteria were developed so that in a medically fit patient the final decision on discharge is made by the nursing and physiotherapy staff. Patients should be independent for transfers in and out of bed, have managed stairs and showering, and have satisfactory knee flexion (for knee replacement patients) prior to discharge. The ERAS team consulted and discussed the new protocols with all orthopaedic surgeons, junior doctors, anesthetists, nursing staff, inpatient and outpatient physiotherapists, and local GPs.
Following discussion and small-scale trials, the new protocols were implemented in January 2013. Key aspects are summarized in Table 2.
This study compares a consecutive series of all patients undergoing primary elective total hip (THR) or knee replacement (TKR) from January 1, 2013, to June 30, 2014, with a historical control cohort from the preceding 18-month period, July 1, 2011, to December 31, 2012. Revision surgery and hip replacement for acute fractures were excluded. Data including age, gender, body mass index (BMI), ALOS, time of discharge, transfusion requirements, and acute readmissions were collected prospectively.
Baseline data collected included the American Society of Anesthesiologists (ASA) grade. This has five grades, where ASA I is a healthy patient, ASA II is a patient with mild systemic disease, ASA III is a patient with severe systemic disease, and ASA IV is a patient with severe systemic disease that is a constant threat to life. ASA V is a moribund patient who is not expected to survive without the operation and is not relevant to elective joint replacement. (ASA, 2014) Preoperative Oxford hip or knee scores (OHS, OKS), which are used in our prioritization process, were also collected for the ERAS group. The Oxford score is a 12-question patient-reported outcome measure. There are five questions on pain and seven on function, each of which has five options and is scored 0-4, where 0 is worst and 4 is best. This gives a score of 0-48, where 48 is best. The score is validated and widely used to report the improvement and outcome following hip or knee replacement surgery (Murray et al., 2007).
Complications and readmission data were collected from our institution's patient management system, the department's surgical audit system, and the surgical site infection (SSI) surveillance program for hip and knee replacements. Data were cross-referenced with the New Zealand (NZ) Joint Registry for revisions, deaths, and 6-month postoperative Oxford scores. The NZ Joint Registry (2014) collects details of all patients undergoing hip or knee replacement in the country and has 98% compliance.
The study was approved by the University of Otago Human Ethics committee (Health). Statistical analysis was performed using Stata v13 (College station, TX). Two-tailed t tests were used for continuous variables and chi-square tests for categorical data.
Results
There were 507 patients who underwent primary elective hip and knee replacements in the historical control group and 528 in the ERAS group. The groups were well matched with respect to age, gender, and ASA grade (see Table 3). The mean age for patients who underwent hip replacement was 68 years and for patients who underwent knee replacement was 70 years. In both groups, 30% of patients were ASA Grades III and IV, indicating the presence of severe systemic comorbidities. In the ERAS group, the median BMI was 31.7 kg/m2 and the mean preoperative OHS was 11.1 (SD = 4.9) and OKS was 11.1 (SD = 4.0). These data were not available for the historical cohort but are likely to be very similar.
The day of surgery admission rate increased from 96% in the control group to 99% in the ERAS group. The average ALOS for elective THR fell significantly by 1.3 days and by 0.9 days for TKR, with the main drop occurring almost immediately following the introduction of the new protocols, especially for THR (see Figure 1). There was a shift to discharges later in the day during the course of the study. If patients were able to be discharged in the afternoon or evening after their physiotherapy, they did not have to wait for a medical ward round the following day. Patients who were mobilized on the day of surgery had an ALOS 1 day shorter than those who did not. Twenty-one patients (4%) were discharged to a rehabilitation ward or rural hospital bed. The mean age of these patients was 76 years, and 17 (81%) were patients who underwent hip replacement.
Following the introduction of the perioperative anemia management pathway, the transfusion rate dropped from 26% to 17% for patients who underwent hip replacement (p = .18) and was unchanged for patients who underwent knee replacement at 9%. There was a 40% reduction in the number of units of blood transfused per patient. The median length of stay (LOS) for patients with preoperative anemia was 5 days compared with 4 days for nonanemic patients.
Two in-hospital deaths occurred in ASA Grade III patients. A 78-year-old died of aspiration pneumonia after prolonged hypotension following a hip replacement. An 80-year-old developed cardiogenic shock following knee replacement and died despite acute coronary artery bypass grafting. Two patients (aged 86 and 79 years, ASA Grade IV with multiple comorbidities) died following discharge to residential care at 30 and 50 days, respectively. The 30-day and 90-day death rates were not significantly different from those of the historical cohort (see Table 3).
There were five confirmed SSIs out of 528 hip and knee replacements (0.9%). Four patients (0.75%) required a return to the operating theater: two patients who underwent knee replacement for superficial infection and wound breakdown, and one patient for a hematoma following hip replacement. A 78-year-old man required revision of an uncemented hip at 4 days due to an early dislocation because of loose undersized components. Three other patients who underwent hip replacement had revision at 28, 133, and 455 days for dislocation, giving a hip revision rate of 1.3%. No patients who underwent knee replacement had revision. The revision rates were not significantly different from those seen in the historical cohort.
There was an increase in the 30-day readmission rate from 3.2% (16 of 507) to 5.5% (29 of 528) (ns, p = .065). In the ERAS group, the commonest reasons for readmission were for wound problems and suspected infection (n = 9), pain issues (n = 7), and concerns regarding leg swelling and deep vein thrombosis (n = 4). Only four patients from the historical group were readmitted for these reasons.
Regular qualitative patient surveys were performed, with high satisfaction reported. The 6-month Oxford scores were higher in the ERAS group by 3.5 points for patients who underwent knee replacement (p = .03) and 2.2 points for patients who underwent hip replacement (ns, p = .152) (see Table 2).
Compared with the historical group, there was a saving of 601 bed nights over the 18-month study period or 400 bed nights per year. This represents a theoretical bed night reduction of 20%. However, in practice, this opened up the beds for additional patients with an increase in elective admissions of 10.7% and 6% for acute admissions, resulting in no significant change in the total number of orthopaedic bed nights. The increase in readmissions led to an extra 97 nights (163 nights vs. 68 nights in the control cohort).
Discussion
There has been a trend toward shorter LOS following hip and knee replacements over the last two decades. Day of surgery admission has become the norm in our unit, and only in exceptional circumstances are patients not admitted on the day of surgery regardless of their place of residence, age, or family supports.
Enhanced recovery initiatives have been shown across the world to improve patient outcomes and can reduce the LOS to 2-4 days for patients undergoing hip and knee replacements even in unselected cohort studies (Kehlet, 2013). In practice, many hospitals have longer stays than this and may discharge patients to rehabilitation facilities rather than home. However, in the United Kingdom, it appears that hospital stays of around 5 days are achievable with the widespread implementation of ERPs in public hospitals and not just in dedicated specialist units (Kehlet, 2013).
This study reports on a consecutive unselected group of the first 528 primary hip and knee replacements performed in our public hospital following the implementation of the ERAS program. No patients were excluded. The mean age of 68 years for patients undergoing hip replacement and 70 years for patients undergoing knee replacement was typical for public hospital patients (Dakin, Gray, Fitzpatrick, MacLennan, & Murray, 2012; Jenkins et al., 2013; Malviya et al., 2011; McDonald et al., 2012; Wainwright & Middleton, 2010). The BMI (31.7 kg/m2) was higher than NZ averages for hip (28.7 kg/m2) and knee (31.1 kg/m2) (NZ Joint Registry, 2015). The proportion of patients with ASA Grades III and IV (32.7%) was higher than that reported by Wainwright and Middleton (2010; 8.5%) and the NZ Joint Registry (2015; 23%-25%). In our district, we have a population that is older than the national average and has had problems with access to hip and knee replacements (Gwynne Jones, 2013; Gwynne Jones & Iosua, 2016). Many patients undergoing surgery had been waiting for 12-18 months due to financial constraints. Most patients had severe disease and were significantly deconditioned. This is reflected in the average preoperative Oxford score of 11 points for both hip and knee replacements compared with average preoperative Oxford scores of 18-20 points reported by others (Dakin et al., 2012; Jenkins et al., 2013; McDonald et al., 2012). Despite this, we were able to show significant changes in our LOS, with only a small increase in readmission rates. Our median LOS dropped from 5 to 4 days for patients who underwent knee or hip replacements, and the reduction in mean LOS is comparable with other published results from the United Kingdom (Kotze, Carter, & Scally, 2012; Malviya et al., 2011; McDonald et al., 2012; Robinson, Wagstaff, Sanghera, & Kerry, 2014; Wainwright & Middleton, 2010). Units in the United States have reported decreases in LOS of 3.4-3.5 days with similar strategies including preoperative education classes and pain management (Kearney et al., 2011; Parisien, Valentine, Hoffman, & Penzero, 2012). However, LOS figures can be misleading if some patients are discharged to step-down facilities. Large studies from the United States have shown that, despite a mean LOS of 3.9-4.2 days, 34%-48% of patients are discharged to rehabilitation or skilled nursing facilities (Schairer, Vail, & Bozic, 2014; Zmistowski et al., 2013). Our figure of 96% discharged home is very similar to that of the Bournemouth group in England (Wainwright & Middleton, 2010).
It has been shown previously that transfusion is associated with longer hospital stays and readmission rates (Ibrahim et al., 2013; Kotze et al., 2012) and preoperative hemoglobin level predicts ALOS independent of transfusion (Kotze et al., 2012). In our initial audit, we noted that preoperative anemia was associated with an increase in median LOS of 1.5 days. The increase in median LOS is now 1 day. Our transfusion rate remains higher than that of other studies (Irwin et al., 2013; Kotze et al., 2012; Malviya et al., 2011; McDonald et al., 2012). If this can be further reduced, there are potential benefits both in terms of LOS and in the costs of blood transfusion.
The complication rate has not increased significantly. Our SSI rate of 0.9% is similar to the NZ average of 1.2% (Morris, 2014) and at this stage no deep infections have been identified. The return-to-theater rate of 0.75% compares favorably with reported rates of 1.3%-1.8% (Irwin et al., 2013; Malviya et al., 2011). The 30-day readmission rate increased from 3.2% to 5.5%, but this is still similar to the 28-day readmission rates of 3.1%-8.5% reported by others (Irwin et al., 2013; Malviya et al., 2011; Schairer et al., 2014; Wainwright & Middleton, 2010; Zmistowski et al., 2013). It is concerning that the number of readmissions directly related to the orthopaedic procedure such as wound problems, pain, and swelling has increased. It is not clear whether this is related to the age and comorbidities of the patients or early discharge.
The 30-day and 90-day death rates of 0.57% and 0.76% are a little higher than those reported in other contemporary series (Cusick & Beverland, 2009; Hunt et al., 2013; Malviya et al., 2011; Sharrock, Della Valle, Go, Lyman, & Salvati, 2008). This may reflect the high proportion of patients who were ASA Grades III and IV. Malviya et al. (2011) reported a reduction in mortality rates, with ERPs from 0.5% to 0.1% in an unselected group of patients of a similar age, but did not report ASA grade. Data from the U.K. registry have shown a reduction in 90-day mortality from 0.56% to 0.29% over the 8 years from 2003 to 2011 (Hunt et al., 2013).
Like others, we found patient satisfaction was high (Jones et al., 2014; Kearney et al., 2011; Parisien et al., 2012). The education classes were particularly well received. The 6-month postoperative OKSs of 39.8 in the ERAS group were 3.5 points higher than those of the historical group and higher than the NZ average of 37.4 (NZ Joint Registry, 2015). The minimum clinically important difference in Oxford score has been reported as 5 points but may be as little as 2 points (Beard et al., 2013; Murray et al., 2007). Therefore, patients in the ERAS had at least equivalent outcomes at 6 months to the historical cohort. The early revision rate (1.3% at minimum 18-month follow-up) for patients who underwent hip replacement is comparable with the NZ Joint Registry rate of 1.1% at 1 year and 1.6% at 2 years (NZ Joint Registry, 2015).
A strength of this study is that it reports on a consecutive group of patients presenting to a general public hospital with significant comorbidities and severe symptoms. There were multiple surgeons and anesthetists of varying grade involved in the surgery, and a variety of implants were used. This suggests that the interventions are generalizable and similar results can be achieved in other institutions.
A limitation of our study is the use of a historical control group, although the patients were well matched. During the development of our protocols, we noted a small drop in ALOS. In the first 6 months of the historical cohort period, the ALOS was 5.8 days for patients who underwent hip replacement and 6.1 days for patients who underwent knee replacement. The ALOS had fallen to 5.4 days for patients who underwent hip replacement and 5.5 days for patients who underwent knee replacement by the date we formally went live. The dramatic drop after this date is particularly evident for patients who underwent hip replacement and strongly suggests that the gains were due to the new program. If we took the first 6-month period of the control group as our baseline, then the potential benefits would increase to 500 bed nights per year. This is in part offset by the increased bed nights due to readmissions.
Some of the improvements were small and may not, on their own, have been shown to have a statistically significant effect. However, a key component of programs such as these is the aggregation of marginal gains (Durrand, Batterham, & Danjoux, 2014). The end result of all the small changes has resulted in significant reductions in key outcomes such as LOS.
For a project such as this to succeed, strong clinical input is needed. It proved critical to engage and enthuse the senior nursing and allied health staff on the ward. Most of the changes instituted were nurse or physiotherapist led. The change in attitudes of surgeons and the junior medical staff followed. Many of the changes can be implemented without any increase in budget. A key failing of the project was that there were no mechanisms to return any of the gains to the department for the benefit of orthopaedic patients. Sustainability has been a challenge. Increasing financial pressures, the loss of key staff, medical outliers on the orthopaedic ward, and nursing shortages are constant threats to the continuing success of the program.
Conclusion
Enhanced recovery programs for patients undergoing hip and knee replacements can significantly reduce the LOS without relying on step-down or rehabilitation facilities. They are effective for an unselected public hospital population with severe osteoarthritis and significant comorbidities and have good outcomes, high patient satisfaction, and no increase in complications. Strong nursing, physiotherapy, and clinical leadership and a multidisciplinary approach are required.
Acknowledgments
Terry Wall, Charge Nurse Elective Orthopaedics, and Clare Hargest, physiotherapist, for their work in developing and implementing the new procedures; Dr. Justin Holborrow for developing the anesthetic guidelines; Dr. Jim Faed, consultant hematologist and transfusion medicine specialist, for help in developing the perioperative anemia algorithm; Associate Professor Bob Hancox, for help with statistical analysis; and Messrs Rob Middleton and Tom Wainwright, for their advice and enthusiasm on setting up the program.
References