Development of reference charts for monitoring quadriceps strength with handheld dynamometry after total knee arthroplasty

Abstract

Purpose

To develop reference charts which describe normative quadriceps strength recovery after total knee arthroplasty (TKA) as measured by handheld dynamometry (HHD).

Materials and Methods

We conducted a retrospective analysis of post-TKA quadriceps strength recovery in a longitudinal dataset consisting of both clinical and research HHD data. We created sex-specific models for recovery using Generalized Additive Models for Location, Scale, and Shape. We created reference charts from the models to display the recovery of population centiles over the first six postoperative months.

Results

A total of 588 patient records with 1176 observations were analyzed. Reference charts for both sexes demonstrated a rapid increase in quadriceps strength over the first 60 postoperative days followed by a more gradual increase over the next 120 days. Males appeared to demonstrate faster recovery and greater strength on average compared to females. The quadriceps strength recovery of three female patient records was plotted on the reference chart to illustrate the charts’ potential clinical utility.

Conclusion

These reference charts provide normative data for quadriceps strength recovery after TKA as assessed by HHD. The reference charts may augment clinicians’ ability to monitor and intervene upon quadriceps weakness—a pronounced and debilitating post-TKA impairment—throughout rehabilitation.

Introduction

Total Knee Arthroplasty (TKA) is the most commonly performed inpatient elective surgery in the United States [1], and the volume of TKAs performed annually is projected to increase exponentially in the near future [2]. TKA alleviates disabling osteoarthritic joint pain and improves health-related quality of life [3], but patients typically fail to recover functionally to the level of their healthy peers [4]. Quadriceps weakness after TKA has been linked to reduced functional performance [5–7] and patient dissatisfaction [8]. Furthermore, quadriceps weakness in the geriatric population is associated with increased fall risk [9,10], reduced independence [11,12], and poor health outcomes [12,13]. Quadriceps weakness is the most pronounced in the acute phase after TKA (up to 60% reduction at 1 month) [14,15] and persists for years after surgery [16].

Although quadriceps strength recovery after TKA is closely linked with postoperative function, it can be challenging to monitor clinically. Electromechanical dynamometry is the gold standard for muscle strength assessment, but many clinicians lack adequate training, the necessary time, or the required equipment to use this method. Manual muscle testing is often used clinically to monitor strength due to its convenience, but it lacks the accuracy and responsiveness necessary for monitoring quadriceps strength recovery after TKA [17]. This has led some researchers to advocate for the use of hand-held dynamometry (HHD) for clinical assessment and monitoring of muscle strength [18,19]. HHD is a valid, reliable, and feasible method for measuring isometric quadriceps strength after TKA [18,20].

Psychometric rigor and clinical utility notwithstanding, there is little information available regarding the typical post-TKA recovery of quadriceps strength as measured by HHD. Thus, clinicians are lacking a means of determining whether an individual’s HHD strength measurement is typical or atypical at any particular postoperative timepoint. Therefore, the purpose of this study was to develop reference charts using strength data collected via HHD to describe normative quadriceps strength recovery after TKA. The overall goal is to improve the precision with which clinicians can assess strength recovery following surgery. Use of a reference chart may enhance clinicians’ ability to (1) identify patients demonstrating suboptimal quadriceps strength recovery, (2) tailor strengthening interventions based on individual patient presentation, and (3) objectively monitor quadriceps strength recovery throughout rehabilitation [21–23].

Methods

Description of data sources and preparation

This analysis relied on quadriceps strength data measured by HHD in both research and clinical settings. Research data were collected as part of a previously published observational study investigating swelling after TKA in Denver, Colorado (n = 56) [22]. Clinical data included were collected throughout the course of routine TKA rehabilitation primarily at two ATI Physical Therapy clinics in Greenville, South Carolina (n = 2555). Clinical data were recorded in a quality improvement database for all patients with TKA without selection criteria. Because the clinical data were collected in real-world practice, data collection timepoints were not uniform across participants. However, clinicians attempted to collect data at least on a semi-weekly basis. Variables available for reference chart development in both data sources included patient demographics, anthropometrics, quadriceps strength HHD measurements, and other clinical measures of recovery. Patient records were retained for analysis if sex, age, and date of surgery were known and if at least one postoperative HHD measurement was recorded with the corresponding date of assessment. Records were excluded from analysis for patients under the age of 30, records obtained more than 3 years after TKA, and physiologically implausible observations which were determined to be data entry errors. All patient records were de-identified before being accessed by the research team at the University of Colorado, and all study procedures complied with a non-human subjects research designation (COMIRB #: 18–1246).

Description of HHD assessment of quadriceps strength

Quadriceps strength was measured with a handheld dynamometer (Lafayette Hand-held Dynamometer, Lafayette Instrument Evaluation, Lafayette, Indiana, USA) by licensed physical therapists or physical therapist assistants in both research and clinical settings. HHD assessments were conducted using a technique validated for measuring quadriceps strength after TKA [20]. Briefly, a “make” test of isometric knee extension force was performed as follows: (1) patient is seated with 90$°$ of hip flexion and 60$°$ of knee flexion with hands in lap, (2) clinician places dynamometer approximately 5cm proximal to the medial malleolus, (3) patient exerts a maximal knee extension force while the clinician attempts to match patient’s force with the dynamometer. HHD force values were measured in pounds and subsequently normalized to percentage of the patient’s body weight (HHD force / bodyweight * 100). HHD force was typically measured twice within each assessment period; the average of these values was used for analysis. This assessment technique has excellent test-retest reliability (ICC = 0.91) and responsiveness (standardized response mean >1.2) for patients with TKA [20].

Data Analysis

The statistical models for the reference charts describe the variation of weight-normalized knee extensor strength over the first year following surgery. Therefore, preoperative measurements were not included in this analysis. Measurements recorded between 1–3 years after surgery were all analyzed as 1 year to avoid model overfitting at later postoperative timepoints where data were sparse. For our main analyses, all available data were used. Analyses were stratified by sex due to its known impact on quadriceps strength recovery after TKA [24,25]. All analyses were performed using R statistical software (https://www.R-project.org/) [26]. Briefly, a Box-Cox Cole and Green distribution was used to model changes in the median, variance, and skewness for the outcome (quadriceps strength) over time, via Generalized Additive Models for Location Scale and Shape (GAMLSS) [27], Each of the parameters was flexibly modeled with cubic splines, and the optimal degrees of freedom (knots) for each parameter was determined using the “find.hyper” function in GAMLSS. Quadriceps strength recovery was modeled for the first year after TKA, but only the first six postoperative months were visualized in the reference charts. We used the ggplot2 package to visualize the 5^th, 10^th, 25^th, 50^th, 75^th, 90^th, and 95^th centiles of the model from postoperative days 2–180 [28]. We examined each model’s goodness of fit by calculating the percentage of actual observations which fell below each specified centile in the model (appendix table 1). We then carried out sensitivity analyses by modeling quadriceps strength recovery using a limited dataset which contained only one randomly selected observation per individual to examine the potential influence of serially correlated data. We calculated the absolute differences at each timepoint between 10^th, 50^th, and 90^th centile curves generated using the full and limited datasets; a difference of less than 10% was considered acceptable (appendix figures 1 and 2).

Results

A total of 2611 patient records (5497 observations) with surgical dates spanning from 2013–2020 were assessed for inclusion in this study. After screening and data cleaning, 588 patient records (1176 observations) were available for analysis. A total of 952 observations were preoperative and therefore not included in the analysis, while 1677 records were not useable due to missingness (e.g., missing date of assessment or date of surgery). The final dataset included 345 female records (669 observations) and 243 male records (507 observations). The mean observations per record was 1.9 (SD 1.2) for females and 2.1 (SD 1.2) for males. Figure 1 describes the data excluded based on each selection criterion, data excluded due to missingness, and patient demographics for the final female and male datasets.

Figure 1.

Flow diagram and demographics for data included in final analysis

Reference charts for HHD-assessed quadriceps strength recovery after TKA

Female and male reference charts were developed for quadriceps strength recovery after TKA as measured by HHD from postoperative days 2–180 (figures 2 and 3). These reference charts are available in an interactive R Shiny web application (https://graberje.shinyapps.io/Quad_Strength_TKA/) [29]. The median prediction for female strength (% body weight) improved from approximately 2% to 21% over the first 180 postoperative days. The 10^th centile improved from approximately 0% to 12% and the 90^th centile from 5% to 31%. The median prediction for male strength improved from 2% to 26%. The 10^th centile improved from 0% to 16% and the 90^th centile from 6% to 38%. Males and females demonstrated similar quadriceps strength values at postoperative day 2 across centiles. The first 60 postoperative days for both sexes were characterized by a rapid increase in quadriceps strength followed by a more gradual increase over the next 120 days. However, the slope was steeper across all centiles in the male reference chart, and males demonstrated greater strength on average at every postoperative timepoint.

Figure 2.

Reference chart describing post-TKA quadriceps strength recovery for females.

The centiles for female quadriceps strength recovery (5^th, 10^th, 25^th, 50^th, 75^th, 90^th, and 95^th) are plotted continuously over the first 180 postoperative days.

Figure 3.

Reference chart describing post-TKA quadriceps strength recovery for males.

The centiles for male quadriceps strength recovery (5^th, 10^th, 25^th, 50^th, 75^th, 90^th, and 95^th) are plotted continuously over the first 180 postoperative days.

Examples of individual patient cases

We extracted three individual patient records from our dataset to illustrate the reference charts’ potential for clinical application. Only female cases were chosen to allow each patient trajectory to be plotted and compared on the same reference chart (figure 4). We restricted this chart to the first 60 days following surgery to focus our discussion on the timeframe when quadriceps strength changes most rapidly.

Figure 4.

Reference chart with example female patient cases.

The observed quadriceps strength values are plotted for Patient A (green), Patient B (red), and Patient C (purple) over the first 60 postoperative days.

Patient A (green line) demonstrates one of the better trajectories of postoperative quadriceps strength recovery; the patient is stronger than 80-90% of patients at every measurement point over the first 47 postoperative days. Patient B (red line) demonstrates quadriceps strength in the bottom 10% of patients over the first postoperative week, but their strength improves rapidly (up to the 75th centile by day 44) during the subsequent weeks. Patient C’s quadriceps strength recovery (purple line) looks similar to Patient B’s early recovery; their strength fluctuates between the 25th and 50th centiles during the first two postoperative weeks. However, Patient C’s strength diminishes over the next month placing the patient near the bottom 5th centile at day 45.

Discussion

We developed reference charts to describe strength recovery, as measured by HHD, for the first 180 days following TKA surgery. The charts are not intended to guide a specific intervention strategy but to inform the professional judgment of skilled clinicians. The goal is to provide clinicians and patients a template against which to measure the strength recovery of individual patients, as there is currently a lack of accepted benchmarks for monitoring quadriceps strength recovery after TKA using HHD. Current methods which clinicians may use to gauge quadriceps strength recovery via HHD have notable limitations. For example, comparing the patient’s strength to their contralateral limb is problematic as contralateral limb strength is often impaired in knee osteoarthritis [30]. Monitoring strength relative to population norms may also have limited utility in the early postoperative period when dramatic strength impairments are expected [14,15]. Therefore, the reference charts created in this study may provide increased clinical value to the objective measurement of quadriceps strength with HHD. In particular, the charts can be used to inform patient and clinician expectations of postoperative recovery, as well as judgments of treatment success and failure throughout rehabilitation.

The example cases present three plausible clinical scenarios where the reference charts could be useful. Although each patient demonstrates quadriceps weakness throughout the observed period, the extent of each patient’s weakness relative to expected can be used to inform care decisions. Patient A’s strength is consistently higher than most; the patient might be a candidate for reduced care frequency or earlier discharge planning depending on their recovery in other outcomes (e.g., pain, function). Patient B’s rapid strength improvement could provide evidence for maintaining their current treatment plan. Conversely, Patient C’s strength decline suggests the patient may not be responding to their current strengthening regimen. With use of the reference chart, the clinician might identify and intervene upon this suboptimal response more quickly. Examples of how the clinician could choose to intervene include (1) altering treatment mode or intensity, (2) increasing rehabilitation frequency, or (3) investigating external factors which might be influencing strength recovery. The reference chart’s utility in each of these cases depends on the patient’s overall clinical presentation, the patient’s preferences, and the clinician’s professional judgment. However, these simple examples illustrate how the charts could provide increased meaning to HHD quadriceps strength assessments. We envision this may improve clinicians’ ability to intervene upon quadriceps weakness and, therefore, facilitate an increased emphasis on restoring quadriceps strength during TKA rehabilitation.

Although these reference charts describe normative quadriceps strength recovery for patients with TKA, they do not provide strength information relative to healthy population norms. This has important clinical implications as most patients do not recover quadriceps strength to the level of their healthy peers after TKA [4,16]. For example, Patient A demonstrated quadriceps strength in the 90th centile at each measurement time point relative to females with TKA, but their last strength measurement is still 30% lower than population norms for females their age (60–69) [31]. Clinicians seeking to use our charts—or any quadriceps strength monitoring strategy—to inform clinical decision making should be aware that even top-performing patients are likely to have impaired quadriceps strength after TKA relative to the general population.

The overall quadriceps strength recovery trajectories in our reference charts are comparable to previous work conducted with different strength assessment techniques [5,25,32]. This includes our observation that males appeared to demonstrate greater strength and a more rapid recovery than females [25,32]. The reference charts developed by Pua and colleagues in Singapore are the closest comparison for our results [32]. Pua et al. measured quadriceps strength isoinertially on a seated knee extension machine instead of isometrically via HHD but had similar findings. Both our reference chart and Pua’s reference chart for females appear to demonstrate similar trajectories and close agreement across centiles over time. The charts for males also look similar, but the slope and relative strength values appear to be greater across centiles in the chart developed by Pua et al. This difference could be explained by the tendency for HHD to underestimate force output at higher force levels compared to alternative strength assessment techniques [19,20,33,34]. Overall, the similarity between study results supports the generalizability and potential utility of our findings for clinical practice.

Limitations

As mentioned above, the primary limitation to using HHD for quadriceps strength assessment is its tendency to underestimate force output when the tester is weak and/or the patient is strong [19,20,33,34]. This tendency may not be particularly relevant for monitoring quadriceps strength after TKA in the early postoperative period when most patients are quite weak. Furthermore, we believe the primary utility of HHD and our reference charts in TKA recovery is to identify weak individuals—especially early after surgery when clinical interventions may be most effective [14,35,36].

There are several limitations to our study design. The patient records used to develop our reference charts were gathered from only a few clinical and research sites. The generalizability of our findings may be limited by (1) possible selection bias related to data missingness, (2) differences in patient demographics and characteristics at different clinical locations, and (3) potential changes in TKA-related care during our data collection period (2013–2020). However, the similarity between our findings and research conducted in Singapore suggest our reference charts may be generalizable to populations outside of our sample [32]. We also contend that using primarily clinically collected data may improve the generalizability of our findings as no restrictive selection criteria were used for data collection. The non-uniform timepoints in the clinical data also allowed us to better capture the trajectory of postoperative quadriceps strength recovery.

Conclusion

The reference charts developed in this analysis may increase the clinical utility of HHD—an objective and feasible technique—for monitoring post-TKA quadriceps strength recovery. Patients demonstrating suboptimal quadriceps strength recovery may be more quickly identified, providing clinicians with the opportunity to adjust treatment strategies accordingly and prospectively monitor patients’ response to treatment. Additionally, these charts may increase the value of HHD for goal setting and plan of care decision making throughout TKA rehabilitation.

Appendix

Appendix Table 1.

Examining model fit for male and female reference charts. Values are displayed for the percentage of observations which fell below each model centile. Ideally, 50% of cases would fall below the 50^th centile, 75% of cases would fall below the 75^th centile, etc.

Model Centile	% of observations below (F)	% of observations below (M)
5th	4.6	3.7
10th	10.8	8.9
25th	25.6	25.6
50th	49.5	51.3
75th	74.4	73.4
90th	89.2	89.5
95th	94.5	95.7

Appendix Figure 1.

Female reference chart sensitivity analysis for repeated measures.

The full model (black) is plotted against the limited model with only one measurement per patient (light blue). The models demonstrate close agreement across centiles which suggests the full model is robust to repeated measurements. The mean difference between the models across timepoints is 0.09% bodyweight for the 10^th centiles, 0.23% bodyweight for the 50^th centiles, and 0.30% bodyweight for the 90^th centiles.

Appendix Figure 2.

Male reference chart sensitivity analysis for repeated measures.

The full model (black) is plotted against the limited model with only one measurement per patient (light blue). The models demonstrate close agreement across centiles which suggests the full model is robust to repeated measurements. The mean difference between the models across timepoints is 0.08% bodyweight for the 10^th centiles, 0.28% bodyweight for the 50^th centiles, and 0.37% bodyweight for the 90^th centiles.