What Is the Learning Curve for New Technologies in Total Joint Arthroplasty? A Review

Abstract

Purpose of Review

The adaptation of new technology in joint replacement surgery is often associated with a learning curve, as performance tends to improve with experience. The purpose of this review is to define the learning curve and its relevance to joint replacement surgery in the setting of new technological advances, and to draw analogies with the learning curve of basic surgical training.

Recent Findings

Assessing a surgeon’s learning curve for a new technology is complicated and difficult. With every learning curve, the first patients subjected to the novel technology may be at higher risk for adverse events until the learning curve is overcome and a steady state is reached. While measures of performance can be clear and direct in some professions, learning curves with new technology in total joint arthroplasty have been difficult to quantify. Most attempts measure surgical learning curves via an evaluation of the surgical process or patient outcomes. There are published results of both process (i.e., operative time, accuracy of implant position) and outcome measures (i.e., complication rate, revision rate) utilized as proxy for performance during learning curves.

Summary

We review the concept of the learning curve in joint replacement surgery, highlighting examples of learning curves with adaptation of new technologies, and conclude with a discussion of dilemmas and challenges.

Introduction

When learning to use a new technology, performance often improves with experience. This concept of a learning curve with new technology was introduced in 1936 by T.P. Wright, an aeronautical engineer who reported that an increase in the experience and skill of the workforce correlated with increased efficiency of aircraft component production and decreased costs [1]. While measures of performance are often straightforward in industry, learning curves and performance with new technology in total joint arthroplasty have been difficult to measure.

The learning curve with new surgical techniques and technologies in total joint arthroplasty may be separated into two distinct measures—surgical process or patient outcomes [2]. Unfortunately, while patient outcomes are a reasonable proxy for assessment of the technology and its learning curve, they often lend themselves to dichotomous events (i.e., complications, survival) which make quantification of the learning curve difficult. More commonly used measures involve the surgical process, including operative time, blood loss, movement of instruments, and the number of “near misses.” The cumulative sum (CUSUM) analysis has been used for quality control in several industries and quantifies the learning curve of a surgical procedure or new technology based on outcome of cumulative performances with a standard reference [3]. It has been used increasingly to assess the learning curves in medical procedures and surgeon performance [4].

Presently, numerous reports examine the learning curve during the adoption of various joint reconstruction procedures and technologies [5•, 6, 7•]. It is ethically responsible for arthroplasty surgeons to assess the effects of these learning curves on patient safety and surgical outcomes when considering a new technology or technique. Furthermore, the surgeon should be mindful of ways to optimize the learning curve to minimize risk to the patient. Thus, the purpose of this paper is to define and review the learning curve for new technologies utilized in joint replacement surgery.

Why Are Learning Curves Relevant in Joint Replacement Surgery?

Surgeons continually evaluate new technology and choose whether to implement them as part of the evolution of the surgeons’ practice. For arthroplasty surgery and other similar fields, learning is often complex and encompasses multiple factors: surgical skill, experience, and volume of cases. Generally, improvements are noted rapidly at first, until a steady state is achieved, in a sigmoid-shaped curve. Assessing a specific surgeon’s learning curve for a specific procedure or new technology is complicated and difficult [8]. As a surgeon or a department institutes a new technique or technology, if there is a learning curve, the early exposure patients may be at higher risk for complications or adverse events until the learning curve is overcome and a steady state of complications is reached.

A well-studied example of a learning curve in joint replacement surgery is highlighted in the direct anterior approach (DAA) to total hip arthroplasty (THA). This approach has gained popularity in North America over the past few decades and has been shown to reduce hospital length of stay increase rates of discharge home [9–11] and improved early functional recovery [9, 12, 13]. Although there is also literature describing downsides, many surgeons actively advertise the DAA [14]. Due to these facts, some surgeons who are comfortable with alternative approaches to THA have transitioned to the DAA and experienced difficulty with the transition early on. Published data on the learning curve of transitioning to the DAA THA ranges from 20 to 50 cases [15–20], and for departments transitioning together, the number is cited as between 200 and 300 [8]. It is useful to know at what number of cases the steady state can be reached, but surgeons must evaluate the outcomes of the early patients and the higher chances for adverse events in that cohort to determine whether the increased risk to those patients justifies the transition at all.

In one study evaluating the learning curve of the DAA, the first 46 cases were associated with twice the operative time and estimated blood loss (EBL) as compared with patients undergoing THA via the surgeons’ former approach (posterolateral) [20], with no improvement occurring over the course of those 46 patients. The authors of that study subsequently abandoned the DAA based on their results. Although this study is limited due to the fact that their learning curve may have been greater than 46 patients, it raises important ethical questions—how many patients or how large a learning curve is acceptable, or how great an improvement with a new technology is needed to justify transitioning practice? This stresses the importance of constant vigilance and assessment of patient outcomes when adopting new techniques or technology.

Additionally, in the modern era of medicine, cost must be considered. It is estimated that by 2030, 635,000 THAs will be performed in the USA [21]. As the volume of primary THA continues to rise, so too will revision THA unless implant survivorship improves. Costs for adverse events requiring revision arthroplasty are staggering, with each hospitalization costing up to $31,000 [22]. Another recent study showed that Medicare beneficiaries having an adverse event related to THA had a significantly higher hospital cost [23]. As new technology is adopted and surgeons transition to its use, the early patients in that surgeon’s learning curve may be at increased risk for adverse event or requiring revision surgery. This has the potential to be a heavy economic burden to the healthcare system.

Examples of Learning Curves in Joint Replacement Surgery

Learning curves vary with different types of procedures and new surgical technologies, and often are major barriers to adoption [24]. The learning curve of a new technology is a function not only of the operator but also of the infrastructure surrounding the operator such as surgical assistants, surgical technicians, and the hospital system itself. Further complicating learning curves is that after initial introduction of technology there is continued modification and enhancement of the processes and the device itself. Malcolm Gladwell in Outliers defines the “10,000 rule” as 10,000 h of correct, dedicated practice, as the key to achieving expertise in any field [25]. While there may be many different pressures for a hospital or surgeon to adopt new technologies, patient outcomes should always remain paramount. All of these variables must be considered when evaluating the decision to adopt a new technology in order to ensure safety and satisfactory clinical results.

There are many studies that have evaluated learning curves for various techniques and technologies within arthroplasty surgery and orthopedics. Learning curves have been formulated off of many different variables; however, generally they are a function of operative time and/or complication rate. In a retrospective cohort study by de Steiger et al., the learning curve of DAA THA was based on the cumulative percent revision rates at 4 years of 68 surgeons. They reported that the overall revision rate of DAA THA at 4 years postoperatively was 3%. Surgeons who had performed < 15 procedures at 4 years had revision rates of 6%, compared with 2% for those surgeons whom had performed > 100 operations (p < .05). They observed that it was not until surgeons performed 50 cases that their revision rates approached that of surgeons who had performed greater than 100 cases [17]. Therefore, based on cumulative revision rates, they defined the learning curve of DAA THA to be 50 cases.

Another study reported on the learning curve by operative time with the use of an electronic sensor (VERASENSE [OrthoSensor, Inc., Dania, FL]) to assess soft tissue balancing in total knee arthroplasty (TKA) in a retrospective manner [7•]. The study cohort comprised of 7 groups based on chronicity of 287 consecutive TKA from a single surgeon using this novel technology. Manually balanced knees were used as the control group. They observed that after 41 cases, there was a statically significant decrease in mean operative time (120.4 min for first 41 cases vs. 108.5 min for second 41 cases, p = .0089). Additionally, there was no statically significant difference of operative time between the second group of 41 cases and manually balanced control group (108.5 vs. 109 min, p = .94). Thus, they described the learning curve for this soft tissue-balancing sensor to be < 50 cases [7•].

Lastly, a prospective cohort study by Kayani et al. assessed the learning curve based on not only operative time but also surgeon confidence level, accuracy of implant positioning, limb alignment, and post-operative complication following robotic-arm-assisted unicompartmental knee arthroplasty (UKA) [26•]. In this study they compared 60 conventional jig-based UKAs with 60 robotic-arm-assisted UKAs. They found that there was a sharp inflection point in operative time after the sixth case. They define the initial 6 cases (learning period) to have statically significant longer operative times (p < .001) as compared with the 7th–60th cases (proficiency period). They also found that surgeon confidence was significantly higher in the proficiency period as compared with the learning period. However, they found no difference in radiographic accuracy of implant positioning, limb alignment, or post-operative complications [26•].

Learning curves of new technologies and devices in joint replacement surgery are a barrier to adoption as they impede surgical workflow to the operator and his/her assistants. This temporary obstruction to flow must be carefully balanced when new technologies are under consideration to ensure safe care delivery and positive patient clinical outcomes. Defining learning curves so that a potential adopter can understand and predict how long it may take for him/her to become proficient with using the technology is very helpful when determining whether or not to use new technology. Continued research of learning curves associated with new orthopedic technologies may better guide the decision making process of device adoption.

Barriers and Challenges

The Universal Dilemma

As outlined in an American Academy of Orthopaedic Surgeons’ (AAOS) position statement in 2015, new devices, biologics, and surgical procedures are being introduced and marketed at seemingly ever-increasing rates. While such new technologies may have the prerequisite level of preclinical and/or early and often low-level clinical data to support their introduction, rarely are new innovations supported by sufficient medium- or long-term outcomes data adequate to support their continued safety and efficacy. This gap between the limits of preclinical or early stage clinical results and the actual results in human patients after long-term, large-scale implementation of a new technology has been described by some as the “universal dilemma” [27].

Ultimately, the decision of when to implement the use of a new technology into clinical practice rests with the individual surgeon. Just because a new technology has evidence to support its safe use by some surgeons does not necessarily mean it is ready for widespread implementation by all surgeons, and certainly not without individual surgeons’ receiving adequate training and familiarization before exposing patients to undue risk [28]. The drive to be on the cutting edge of new technologies in the hopes of better outcomes for our patients must be weighed against the risk profile to the patient, the surgeon’s comfort level and familiarity with the new technology, and the available data, often on a case-by-case basis [29].

Challenges to the Patient/Doctor Relationship

The universal dilemma poses ethical challenges to the surgeon wishing to implement new technologies into his/her practice [28], even after they have decided it is worthy of implementation based on the available evidence and training. New technologies run the gamut of minor innovations, such as introduction of a new arthroscope with higher definition cameras, to major innovations, such as introduction of a completely novel surgical technique or implant system. Minor innovations such as the new arthroscope are unlikely to have a significant effect on the patient’s outcome and may not rise to the level of concern to require disclosure. However, a new and relatively untested implant may have significant unforeseen consequences to an early exposure patient, and its use should likely be disclosed to the patient ahead of time. The ethical burden of deciding when and how to disclose the use of new technologies falls on the shoulders of the surgeon.

Surveillance and Follow-up Studies

Collecting and evaluating data on the safety and efficacy of new technologies is an additional challenge to the surgeon on the forefront of clinical implementation as new technologies accrue longer-term, wider-scale use [30]. This responsibility falls on regulatory bodies, industry, and the surgeon. Follow-up studies with proper scientific design, statistical power, and randomization are critical, but remain challenging due to a number of factors including the time, costs, and effort required to conduct such studies. Furthermore, these critical studies are often plagued by low rates of participation due to surgeon or patient preference for one technique or implant over another, and patients are often lost to long-term follow-up, regardless of whether or not new technologies were incorporated into their care.

Conclusion

As new technologies continue to evolve in joint replacement surgery, the surgeon may have to learn and integrate these new technologies into their practice. With the integration of any new technology, there is a learning curve and it is important for surgeons to understand what the reported learning curve means for them and their patients. Furthermore, optimization of the learning curve is key to ensure progress while minimizing patient risk. With experience, the arthroplasty surgeon will achieve a level of proficiency in utilizing new technologies in their clinical practice.

Compliance with Ethical Standards

Conflict of Interest

Nana Sarpong, Carl Herndon, Michael Held, Alexander Neuwirth, and Thomas Hickernell declare that they have no conflicts of interest.

Jeffrey Geller is a paid consultant for Smith & Nephew and receives research support from OSRF.

H. John Cooper is a paid consultant for DePuy Synthes, KCI, Zimmer-Biomet, OnPoint Knee, and Smith & Nephew.

Roshan Shah is a paid consultant for Link Orthopaedics, unpaid consultant for OnPoint Knee, and receives research support from KCI.

Human and Animal Rights and Informed Consent

This article does not contain any studies with human or animal subjects performed by any of the authors.