As hospitals are facing mounting financial pressures in the current economic environment, time spent in the operating room has been identified as one of the most costly areas of hospital operations. As such, introduction of a new total knee arthroplasty (TKA) system to clinical care should demonstrate a minimum learning effort requirement.
To date, limited studies have assessed the learning of new surgical technology or TKA systems. The methodology applied in existing studies usually compares surgical time between the cases performed during the “learning period” and those from the later cases, with an assumed duration (number of cases) of the learning period. In a study of computer assisted TKA, researchers have performed logarithmic regression on the initial case series to find the duration of the learning phase. However, as the surgical time data is often, by nature inconsistent, the regression result can be difficult to evaluate.
Cumulative sum control chart (CUSUM) has been widely applied in industry to assess the stabilization of a production process and has proven to be an objective and effective tool to evaluate a learning process. Although many successes have been achieved by this method in other medical fields, its usage for orthopaedic applications, notably TKA research, has been limited. The goal of this study was to leverage this advanced methodology and perform a CUSUM analysis to define the learning period of a newly released TKA system.
MATERIALS AND METHODS
With institutional review board approval and waiver of informed consent, a retrospective review was performed on the surgical time from four orthopedic surgeons (A-D) on their first 50 consecutive cases since the adoption of a new TKA system, as well as their last 10 cases using their previously mastered TKA system performed before the adoption (baseline). For each surgeon, tourniquet time was used as the primary time measure; while if a surgeon did not routinely use tourniquet, the skin-to-skin time was reviewed instead. Since CUSUM assessed each individual surgeon’s learning process independently, the time measure differences between surgeons did not affect the analysis of an individual’s learning curve as a consistent time measure was used across all 50 cases and baseline for a given surgeon.
To perform the CUSUM analysis, four parameters must be defined (Figure 1A): acceptable failure rate (p0), unacceptable failure rate (p1), type I error rate (α), and type II error rate (β). From the parameters, two decision limits (h0 and h1) and the variable s are calculated. The first 50 cases from each surgeon are sorted chronologically. Each case was evaluated as to whether it “failed” or “succeeded” based on the surgical time criteria defined in Figure 1A. When a failure occurred, a “penalty value” 1-s was added to the CUSUM score; while when a success occurred, a “reward value” s was subtracted from the CUSUM score. A healthy learning process was marked as the CUSUM line crossing the lower decision limit (h0), indicating completion of the learning period (met the acceptable failure rate). Conversely, the CUSUM line crossing the upper decision limit (h1) from below indicated the failure of the learning process (reaching an unacceptable failure rate).
The duration of learning for each surgeon was identified by his/her own CUSUM chart as the number of the last case before crossing the lower decision limit (h0). Surgical time in the baseline, during the learning period and after learning (cases #41-#50) were compared. Significance was defined as p<0.05.
All CUSUM lines from the four surgeons crossed the lower decision limit, indicating their successful completion of learning (Figure1B). The duration of learning was on average 8.3 ± 3.8 cases with individual surgeons exhibiting unique learning characteristics, reflected by the shape of the CUSUM line. Surgeons A and C exhibited significant but moderate time decreases from the learning period to after learning (Figure 2). For all four surgeons, the learning period did not significantly increase their surgical time from the baseline, and the surgical time after learning showed a general trend of smaller standard deviations and shorter time compared to the baseline (Figure 2).
This study applied the CUSUM method to analyze the learning curve of a new TKA system based on surgical efficiency (time), relating the adoption of the surgery as a process that eventually stabilizes with mastery of the task. The data indicated that the learning of the new TKA system took approximately 8 cases. Cases performed, using the new TKA system remained time neutral with cases baseline both during and after the learning period. The data also demonstrated that learning the new TKA system did not result in a significant learning curve from the perspective of surgical efficiency.
Despite the CUSUM method being proposed in the 1970s for analyzing the learning curve for surgical procedures and since then being applied to various medical fields, the use of this method in TKA has been very limited. Utilization of this advanced method in studying the learning curve, not only can provide improved understanding of TKA learning in general, but also allows differences in learning between individual surgeons or surgeon characteristics to be explored.