Predictors of Manipulation Under Anesthesia After Knee Arthroplasty in the United States Military Health System

Abstract

Background

Stiffness after knee arthroplasty is commonly treated with manipulation under anesthesia (MUA). Previously identified predictors of MUA include younger age, Black race, low body mass index, diabetes, tobacco use, and prior nonarthroplasty knee surgery. Whether these factors persist in a healthcare system with minimal financial barriers remains unclear. We evaluated rates and predictors of MUA in the U.S. Military Health System (MHS).

Methods

This retrospective cohort study queried the MHS Data Repository for patients undergoing primary total knee arthroplasty, revision total knee arthroplasty (rTKA), or unicompartmental knee arthroplasty (UKA) from 2016 to 2022 using Current Procedural Terminology. Minimum follow-up was 6 months. Logistic regression evaluated associations between patient factors and MUA.

Results

A total of 8223 knee arthroplasties were identified in 7164 patients. Mean age was 62 ± 9 years; active-duty patients (n = 502) averaged 48 ± 7 years. MUA rates were 4.6% after TKA (315/6833), 2.1% after rTKA (19/885), and 1.4% after UKA (7/505). Protective factors included older age (odds ratio [OR] 0.96/year, P < .0001), UKA (OR 0.23, P = .0002), and rTKA (OR 0.41, P = .0002). Risk factors included Black race (OR 1.90, P < .0001), prior nonarthroplasty knee surgery (OR 1.56, P = .04), and diabetes (OR 1.40, P = .01). Tobacco use and active-duty status were not significant predictors. Mean follow-up was 3 years. No statistically significant association between MUA and revision arthroplasty was detected.

Conclusions

Previously identified risk factors, including younger age, Black race, diabetes, and prior knee surgery, remained associated with MUA within the MHS. Active-duty status was not associated with increased MUA risk.

Introduction

Despite the overall success of knee arthroplasty, some patients experience postoperative stiffness, with limited range of motion (ROM) leading to decreased patient satisfaction and function, and increased costs [[1], [2], [3], [4], [5], [6]]. Manipulation under anesthesia (MUA) can be used to address postoperative stiffness, with reported rates of 0.6% to 10.8% after total knee arthroplasty (TKA) [[7], [8], [9], [10], [11]].

Prior studies have evaluated risk factors associated with MUA in civilian populations. In a study of 229,420 patients, Pfefferle et al. [8] reported an overall incidence of MUA of 1.5%, with risk factors including Black race, female gender, younger age, obesity, and nicotine/tobacco dependence, with younger age and Black race as the most impactful individual risk factors. Knapp et al. [9] explored the impact of preoperative comorbid conditions, type of implant, use of cement, and discharge disposition on the MUA rate after TKA, finding younger age to be risk factors, with no difference between cemented and cementless implants. Contrary to Pfefferle et al., they found that increasing body mass index and current or past status as a smoker was protective against MUA. Barbera et al. [12] investigated racial disparities in MUA rates after TKA and found Black patients to have higher odds of undergoing MUA and longer median time from index TKA to MUA than White patients, suggesting a possible role of racial healthcare disparities in the incidence of MUA.

Active-duty military personnel are generally a young, active, and high-demand population. This population experiences a disproportionately high baseline burden of knee trauma, with a significantly elevated incidence of anterior cruciate ligament injuries [13] and subsequent post-traumatic osteoarthritis [14] compared to the general public. Because prior knee trauma and nonarthroplasty surgical interventions are established predisposing factors for arthrofibrosis, this unique cohort may possess an inherently higher risk for postoperative stiffness. Furthermore, expectation management is a critical component of successful knee arthroplasty, as TKA patients do not always achieve forgotten joint status, defined as little to no awareness of the prosthetic joint during daily activities, postoperatively even though the large majority attain a patient acceptable symptom state [15]. As the Military Health System (MHS) consists of patients of various military or nonmilitary status (active-duty, retired, dependent), we hypothesized that active-duty patients undergoing knee arthroplasty would have higher rates of MUA given their higher activity level and postoperative expectations.

The MHS is a worldwide military healthcare system with relatively uniform insurance coverage and minimal financial barriers to care for eligible beneficiaries, including active-duty service members, retirees, and dependents. It is unknown whether race, obesity, age, diabetes, tobacco use, and prior nonarthroplasty knee surgery remain risk factors for MUA when barriers to accessing care are reduced in comparison to US civilian healthcare, and whether active-duty status confers a higher rate of MUA. We hypothesized that within this healthcare system with reduced financial barriers to care, active-duty status as a proxy for increased patient activity level and expectations and previously published risk factors would predict MUA after knee arthroplasty.

Material and methods

Study design

This was a retrospective cohort study performed to identify risk factors associated with MUA after knee arthroplasty using the MHS Data Repository as a data source. This study was determined to be exempt by our Institutional Review Board.

Study patients

The MHS Data Repository was retrospectively queried for all patients who underwent knee arthroplasty at military treatment facilities from 2016 to 2022 based on Current Procedural Terminology (CPT). Only patients with at least 6 months of follow-up from the date of their captured primary total (TKA; CPT 27447), revision total (rTKA; CPTs 27486, 27487), or unicompartmental (UKA; CPT 27446) knee arthroplasties were included. As some patients underwent bilateral or multiple arthroplasties during the study period, each arthroplasty was analyzed as a separate observation. The proportion of patients contributing multiple procedures was small; therefore, analyses were performed at the procedure level without adjustment for within-patient clustering. Patients included in this study had a mean follow-up of 3 (±1.6) years at the time of data collection.

Risk factors and outcomes

Demographic, medical, and operative risk factors likely to be predictive of MUA based on prior literature were included for each patient. Demographic risk factor data, including race, were obtained from the queried health record database. Race was analyzed as a categorical variable with three groups: White, Black, and Other. Medical risk factor data were obtained using International Classification of Diseases, tenth revision (ICD-10) codes for the following: diabetes mellitus (ICD-10 E10x, E11x, E13x), nicotine/tobacco dependence (ICD-10 F17x), and obesity (ICD-10 E66x) (Appendix Section 3). Prior knee surgery was identified using CPT codes (Appendix Section 2). All study patients were then queried for subsequent MUA (CPT 27570) and further arthroplasty procedures. Procedure, MUA, and revision laterality was confirmed using associated ICD-10 diagnoses. Time to MUA and/or revision was determined directly from the dates associated with the CPT codes for the index procedures and the MUA and/or subsequent arthroplasty procedure.

Statistical analysis

Logistic regressions using generalized linear models were used to measure the effect of patient demographic and comorbidity variables on rates of MUA. Multivariable logistic regression models were used when there were at least 20 events per predictor, otherwise separate univariable logistic regression models were performed for each risk factor. A sensitivity analysis restricted to patients undergoing primary TKA was performed to evaluate the consistency of the primary multivariable model findings. Statistical analysis and data visualization were performed with R v4.2.3 (Posit, Boston, MA) and Microsoft Excel 2016 version 16.75 (Microsoft, Redmond, WA).

Results

Within the MHS, 8223 knee arthroplasties in 7164 patients were performed between 2016 and 2022 (Table 1). Overall average age was 62 ± 9 years, while average age of active-duty patients (n = 502) was 48 ± 7 years (range, 19-70). The MUA rate after TKA was 4.6% (315/6833), performed at a mean of 12 (±5) weeks postoperatively. The MUA rate after rTKA was 2.1% (19/885), at a mean of 10 (±4) weeks. The MUA rate after UKA was 1.4% (7/505), at a mean of 10 (±4) weeks (Fig. 1).

Table 1.

Patient demographics.

Characteristic	MUA (%)	Total
Arthroplasty	341 (4.1)	8223
Surgery
TKA	315 (4.6)	6833
UKA	7 (1.4)	505
rTKA	19 (2.1)	885
Demographics
Active duty	25 (5.4)	462
Age <60	187 (5.9)	3156
Age ≥60	154 (3.0)	5067
White race	157 (3.4)	4572
Black race	113 (6.8)	1668
Other race	71 (3.5)	1983
Prior knee surgery (CPT)	27 (6.8)	396
Diabetes (ICD-10)	87 (4.7)	1847
Nicotine use (ICD-10)	30 (4.5)	672
Obesity (ICD-10)	123 (4.1)	3003

Demographic data of all knees that underwent arthroplasty and had a minimum of 6-month follow-up.

Figure 1.

Schematic of primary statistical analysis. Study flow diagram indicating total number of knee arthroplasties that were evaluated from the MHS data repository and the subsequent primary statistical analysis.

Factors protective against MUA were older age (odds ratio [OR] 0.96/year, 95% confidence interval [CI] 0.94-0.97, P < .0001), UKA (OR 0.23, 0.11-0.49, P = .0002) and rTKA (OR 0.41, 0.25-0.66, P = .0002) (Fig. 2). Obesity approached statistical significance as a protective factor against MUA (OR 0.79, CI 0.63-1.00, P = .052). Risk factors for MUA were Black race (OR 1.90, 1.47-2.44, P < .0001), prior nonarthroplasty knee surgery (OR 1.56, 1.02-2.39, P = .04), and diabetes (OR 1.40, 1.08-1.83, P = .01). Factors that did not significantly impact MUA rate included military active-duty status (OR 0.77, 0.48-1.23, P = .27), nicotine/tobacco dependence (OR 0.96, 0.65-1.42, P = .83), and race other than Black or White (OR 1.06, 0.79-1.41, P = .72).

Figure 2.

Predictors of MUA. Odds ratios with confidence intervals of demographic, medical, and operative risk factors for manipulation under anesthesia following knee arthroplasty (∗P < .05).

A sensitivity analysis restricted to patients undergoing primary TKA demonstrated findings consistent with the primary analysis (Appendix Section 1). Younger age (OR 0.96 per year, 95% CI 0.94–0.97; P < .001), Black race (OR 1.80, 95% CI 1.38–2.35; P < .001), diabetes mellitus (OR 1.43, 95% CI 1.09–1.87; P = .011), and obesity (OR 0.76, 95% CI 0.59–0.97; P = .026) remained independently associated with MUA, whereas active-duty status, prior knee surgery, nicotine use, and other race were not significantly associated.

Of the 8223 knee arthroplasties included, there were 219 revisions (2.7%) during our follow-up period of 3 (±1.6) years [6 mo–6.2 years]. They occurred on average 309 (±329) days after surgery. There were 17 revisions (3.4%) after UKA, 125 revisions (1.8%) after TKA, and 77 revisions (8.7%) after rTKA. Patients with prior knee surgeries and those who were active duty had a higher odds ratio of receiving a UKA as opposed to TKA at index surgery (Table 2). No statistically significant association between MUA and subsequent revision arthroplasty was detected (Table 3).

Table 2.

Factors associated with MUA after UKA, TKA, or rTKA (multivariable logistic regression).

Factor	OR	CI lower	CI upper	P value
Active duty	0.771	0.484	1.228	.273
Age (per year)	0.955	0.942	0.969	<.0001
Non-White, Non-Black race	1.055	0.791	1.408	.715
Black race	1.898	1.474	2.443	<.0001
UKA	0.231	0.108	0.495	<.001
rTKA	0.409	0.255	0.657	<.001
Prior knee surgery	1.563	1.024	2.385	.038
Diabetes	1.409	1.085	1.831	.010
Nicotine	0.959	0.649	1.417	.834
Obesity	0.793	0.627	1.002	.052

Odds ratios with confidence intervals for factors associated with receiving MUA among all patients receiving either UKA, TKA, or rTKA. Bolded values represent statistical significance.

Table 3.

Revision risk after MUA.

Index procedure	OR	CI lower	CI upper	P value
UKA	0.0000007	0	Inf	.99
TKA	1.62	0.82	3.23	.16
rTKA	0.58	0.08	4.39	.60

Odds ratios for likelihood of subsequent rTKA after MUA based on type of preceding arthroplasty.

Discussion

This study investigated risk factors associated with MUA after knee arthroplasty in the MHS, a worldwide healthcare system with minimal financial barriers to care. We sought to determine if previously identified MUA risk factors persist within the MHS and if young, high-demand active-duty patients have an increased risk of MUA.

MUA rates within the MHS were comparable to those reported in civilian arthroplasty populations, suggesting that overall MUA utilization in this system is similar to that reported in other healthcare settings [[7], [8], [9], [10], [11], [12],16,17].

Although the MHS provides relatively uniform healthcare coverage, its arthroplasty population differs from many civilian registries. Beneficiaries include active-duty service members, retirees, and dependents, resulting in a younger population with fewer Medicare-eligible patients and a greater proportion of individuals with prior traumatic knee pathology. Despite these demographic differences, MUA rates and associated risk factors were generally consistent with prior civilian studies, supporting the external validity of these findings while recognizing differences in patient composition.

Patients undergoing UKA and revision TKA demonstrated lower observed rates of MUA than patients undergoing primary TKA. These differences likely reflect variation in patient selection, surgical indications, postoperative ROM expectations, and surgeon thresholds for recommending MUA rather than an intrinsic protective effect of procedure type. Stiffness after rTKA is a common complication. However, goals for ROM after rTKA may be lower than primary TKA and given reconstructive and physiologic limitations, improvement after MUA for rTKA may not be a reasonable expectation [14,[18], [19], [20], [21]]. One study has shown that patients undergoing kinematically aligned TKA, with fewer soft-tissue releases, have lower rates of MUA [11]. The same applies to patients undergoing UKA, who undergo a relatively less invasive surgery with fewer soft-tissue releases, resulting in less perioperative soft-tissue insult, less stiffness, and significantly lower rate of MUA [22].

Consistent with prior literature, we found that Black race, diabetes, younger age, and prior nonarthroplasty knee surgery were factors associated with an increased likelihood of MUA after knee arthroplasty in the MHS. Black race has consistently been reported as a risk factor for postoperative stiffness and MUA in previous studies [8,12,17]. With more equitable access to care, some previously demonstrated racial and socioeconomic disparities in procedures such as colonoscopy screening have been shown to not persist within the MHS [23]. The association observed in the present study is consistent with reports from civilian arthroplasty populations. Okike et al. [24] demonstrated that despite a universally insured population, minority patients had decreased TKA utilization as compared to White patients with Black patients having the greatest disparity. Although Black race remained independently associated with MUA after adjustment for measured covariates, this observational study cannot determine the mechanisms underlying this association. Potential contributors include differences in baseline disease severity, postoperative rehabilitation, ROM, patient preferences, surgeon decision-making, or other unmeasured socioeconomic and structural factors.

Younger age as a risk factor for MUA aligns with most existing literature, as younger patients often have higher functional demands and the surgeon’s decision to proceed with an MUA is impacted by patient functional demands and expectations for knee ROM [8,9,16,18,25]. Although we hypothesized that active-duty patients would have a higher rate of MUA because of their physically demanding occupations and anticipated functional expectations, active-duty status was not independently associated with MUA after adjustment for age and other covariates. Furthermore, a sensitivity analysis restricted to primary TKA demonstrated findings consistent with the primary analysis, supporting the robustness of these results. Together, these findings suggest that younger age, rather than military status itself, is the more important determinant of MUA risk in this population. While active-duty service members are generally younger and more active than other MHS beneficiaries, active-duty status alone does not appear to independently influence the likelihood of postoperative stiffness requiring MUA.

The relationship between MUA after knee arthroplasty and subsequent rTKA has been demonstrated in previous studies. Brigati et al. [10] reported a revision rate of 3.4% at a median time of 9 months after MUA within the American Joint Replacement Registry Medicare population. This revision rate after MUA is approximately twice the mid-term revision rate described by Martin et al. [26] after primary TKA in a similar population. In the present study, we did not detect a statistically significant association between MUA and subsequent revision arthroplasty within the MHS. However, this analysis should be interpreted cautiously given the relatively low number of revision events and variable follow-up duration across patients. The younger age of the MHS cohort and differing indications for revision may also limit direct comparison with predominantly Medicare-based populations.

A primary strength of this study is the use of a comprehensive electronic medical record database within a nationwide healthcare system, providing a large patient cohort for analysis that includes a unique, young, and active-duty military subpopulation. While large database studies evaluating risk factors for MUA are common in the literature, they typically rely on either Medicare data or massive pooled civilian electronic health records, such as the cohort analyzed by Pfefferle et al. [8] These databases carry inherent demographic and financial trade-offs. While Medicare provides a relatively similar model with minimized financial barriers to care, its generalizability is strictly limited to an elderly population (65 years old) with lower baseline functional demands. Conversely, pooled civilian databases capturing younger, working-age populations under commercial insurance may introduce additional variation in access to postoperative rehabilitation because of differences in insurance benefits, out-of-pocket costs, and physical therapy coverage. Because MUA is a highly time-sensitive intervention that relies heavily on early, aggressive rehabilitation, differences in rehabilitation access associated with civilian commercial insurance may influence postoperative care and clinical decision-making regarding MUA. The structure of the MHS provides an opportunity to evaluate these risk factors within a healthcare system with reduced financial barriers to care while including a younger, more active population than is typically represented in Medicare-based datasets.

Limitations of this study include the inherent biases of retrospective administrative database research, including the potential for incomplete or inaccurate coding and documentation. Objective postoperative ROM measurements, patient-reported outcomes, rehabilitation protocols, and surgeon-specific indications or thresholds for MUA were unavailable, limiting assessment of factors directly related to postoperative stiffness and decision-making. Because some patients underwent more than 1 arthroplasty during the study period, analyses were performed at the procedure level rather than the patient level, which may introduce correlation among observations. In addition, variable follow-up duration precluded time-to-event analyses for revision arthroplasty. Finally, although the MHS minimizes many financial barriers to care, other unmeasured socioeconomic and structural factors unique to the military population, including military rank, occupational demands, geographic variability, and cultural attitudes toward pain and return to duty, were not captured and may contribute to the observed associations.

This study adds to the existing body of literature on MUA following knee arthroplasty with the unique lens of minimizing care access differentials in the MHS. Future studies could evaluate patient-specific findings including ROM measurements before and after MUA to evaluate what ROM limitations indicated patients for MUA and whether this differs based on demographic, medical, and surgical factors. Additionally, since active-duty status did not significantly associate with MUA, it may be valuable to explore whether preoperative activity level, as measured with instruments such as the University of California, Los Angeles activity score, associates with MUA after adjusting for the demographic and medical risk factors for MUA validated in this study.

Conclusions

Previously identified risk factors for MUA, including younger age, Black race, diabetes, and prior knee surgery, remained independently associated with MUA within the MHS. Active-duty status was not independently associated with MUA after adjustment. These findings suggest that risk factors identified in civilian populations are reproducible within the MHS, although the mechanisms underlying observed racial differences remain uncertain.

Conflict of interest

Sean E. Slaven receives research support from DePuy. All other authors declare no potential conflicts of interest.

For full disclosure statements refer to https://doi.org/10.1016/j.artd.2026.102124.

CRediT authorship contribution statement

Anthony M. Smyth: Writing – original draft. Ian P. Marshall: Writing – review & editing, Writing – original draft. Jefferson L. Lansford: Writing – review & editing, Supervision, Methodology, Data curation, Conceptualization. Conor F. McCarthy: Writing – review & editing, Supervision, Project administration, Conceptualization. Matthew J. Kinnard: Writing – review & editing, Supervision, Conceptualization. Sean E. Slaven: Writing – review & editing, Supervision, Conceptualization. Robert W. Tracey: Writing – review & editing, Supervision, Conceptualization.

Appendix.

Section 1. Sensitivity analysis: multivariable logistic regression predicting MUA after TKA.

Model D–primary TKA only	OR	CI lower	CI upper	P value
Active duty	0.798	0.491	1.298	0.365
Age (per year)	0.956	0.944	0.969	<0.001
Non-White, Non-Black race	1.043	0.781	1.395	0.779
Black race	1.799	1.377	2.351	<0.001
Prior knee surgery	1.442	0.909	2.286	0.125
Diabetes	1.431	1.089	1.881	0.011
Nicotine	0.963	0.639	1.452	0.844
Obesity	0.756	0.588	0.972	0.029

ORs were estimated using multivariable logistic regression adjusted for all variables shown. The sensitivity analysis was restricted to patients undergoing primary TKA (n = 6833).

Section 2. Prior nonarthroplasty knee surgery CPT codes.

27427, 27427, 27428, 27428, 27457, 27457, 27457, 27705, 27705, 27709, 27709, 29850, 29850, 29850, 29851, 29851, 29851, 29855, 29855, 29855, 29856, 29856, 29856, 29866, 29866, 29866, 29867, 29867, 29868, 29868, 29870, 29870, 29871, 29871, 29873, 29873, 29873, 29874, 29874, 29874, 29875, 29875, 29875, 29876, 29876, 29876, 29877, 29877, 29877, 29879, 29879, 29879, 29880, 29880, 29880, 29881, 29881, 29881, 29882, 29882, 29882, 29883, 29883, 29883, 29884, 29884, 29884, 29885, 29885, 29885, 29886, 29886, 29886, 29887, 29887, 29887, 29888, 29888, 29888, 29889, 29889, 29889.

Section 3. ICD-10 codes.

ICD-10 – Obesity.

E66, E660, E6601, E6609, E661, E662, E663, E668, E669.

ICD-10 – Nicotine.

F17, F172, F1720, F17200, F17201, F17203, F17208, F17209, F1721, F17210, F17211, F17213, F17218, F17219, F1722, F17220, F17221, F17223, F17228, F17229, F1729, F17290, F17291, F17293, F17298, F17299.

ICD10 – Diabetes mellitus.

E10, E101, E1010, E1011, E102, E1021, E1022, E1029, E103, E1031, E10311, E10319, E1039, E104, E1040, E1041, E1042, E1043, E1044, E1049, E105, E1051, E1052, E1059, E106, E1061, E10610, E10618, E1062, E10620, E10621, E10622, E10628, E1063, E10630, E10638, E1064, E10641, E10649, E1065, E1069, E108, E109, E11, E110, E1100, E1101, E111, E1110, E1111, E112, E1121, E1122, E1129, E113, E1131, E11311, E11319, E1139, E114, E1140, E1141, E1142, E1143, E1144, E1149, E115, E1151, E1152, E1159, E116, E1161, E11610, E11618, E1162, E11620, E11621, E11622, E11628, E1163, E11630, E11638, E1164, E11641, E11649, E1165, E1169, E118, E119, E13, E130, E1300, E1301, E131, E1310, E1311, E132, E1321, E1322, E1329, E133.