Secondary Logo

Journal Logo


Socioeconomic Status Is Associated with Risk of Above-knee Amputation After Periprosthetic Joint Infection of the Knee

Lieber, Alexander M. BA; Kirchner, Gregory J. MPH; Kerbel, Yehuda E. MD; Moretti, Vincent M. MD; Vakil, Jeffrey J. MD; Brahmabhatt, Shyam MD

Author Information
Clinical Orthopaedics and Related Research: July 2019 - Volume 477 - Issue 7 - p 1531-1536
doi: 10.1097/CORR.0000000000000634



Above-knee amputation (AKA) is an uncommon but catastrophic complication of TKA. Previous studies have indicated that AKA may be more common in some patient populations than in others. AKA secondary to vascular disease is more common in patients with nonprivate health insurance, female sex, and black race [4,21]. Recent estimates of the frequency of AKAs after TKA range from 14 of 10,000 to 41 of 10,000 with periprosthetic joint infection (PJI) of the knee contributing to most of these cases [11, 16, 19]. Although the rate of AKAs has declined over the past two decades, the rate of AKAs resulting from PJIs, most often associated with TKA, appears to be increasing [7]. Most studies examining AKA specifically after TKA have been case series with limited sample sizes [6, 11, 16, 17, 19]. Given the rarity of AKA, large national databases such as the National Inpatient Sample (NIS) are well suited for the study of this devastating complication. Two recent studies have used the NIS to investigate the prevalence of and factors associated with AKA after TKA [7, 8]. George et al. [7] found that the incidence of AKA due to PJI increased by more than 200% across a 15-year period.

Given the high mortality rate and poor functional outcomes subsequent to AKA after PJI of the knee, further investigation is needed to determine whether this complication is associated with socioeconomic disparities in patients. Notably, previous research has indicated that black race, and age older than 80 years or younger than 50 years were associated with an increased frequency of AKA after TKA [8]. In an analysis of the Medicare 100% National Inpatient Claims Database, Son et al. [20] reported that females with PJI of the knee were less likely to undergo AKA than males. Furthermore, the same study found that there was an increasing risk of AKA after TKA as the number of patient comorbidities increased, but the study found no association between black race and an increased likelihood of AKA. Therefore, the association between race and gender with AKA after PJI of the knee has not been definitively established. Moreover, to our knowledge, prior studies have not addressed whether a patient’s socioeconomic status affects the likelihood of AKA after PJI of the knee.

We therefore asked: (1) Is low socioeconomic status or use of public health insurance plans associated with an increased risk of AKA after PJI of the knee? (2) Is race or sex associated with an increased risk of AKA after PJI of the knee?

Materials and Methods

This retrospective cross-sectional study used the NIS, Healthcare Cost and Utilization Project, Agency for Healthcare Research and Quality [10]. The NIS captures a 20% representative sample of all inpatient discharges from community hospitals in the United States (excluding long-term acute care and rehabilitation hospitals). Data from all insurance types including private, Medicare, and Medicaid are captured in the NIS. We used the NIS between the years 2010 and 2014 based on the availability of the database at our institution. While a database such as the NSQIP would have allowed us to analyze a longer followup period, our research focused on factors associated with AKA incidence not complications or functional outcomes associated with AKA. The NSQIP provides data on 30-day followup, although research indicates that most AKAs after TKA occur 2 to 6 years after the initial surgery [6], so the NSQIP would not have allowed us to more conclusively identify AKA that resulted from PJI of the knee. Additionally, unlike the NIS, the NSQIP would not have allowed us to examine the association between patient income and AKA incidence. The institutional review boards of the institutions participating in this study deemed this research exempt from approval because it used publicly available nonidentifiable data.

We used International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) diagnosis and procedure codes to identify 32,907 patients with a PJI of the knee (see Appendix, Supplemental Digital Content, To be included, patients needed have a diagnosis code related to an infected orthopaedic implant (ICD 996.60, 996.67, 996.69) in addition to a procedure code specific to revision of knee prosthesis, removal of knee prosthesis, and removal or addition of cement spacer (ICD 00.80 00.81, 00.82, 00.83, 00.84, 78.66, 78.67, 84.56, 84.57). Using this approach, we were fairly confident that the patients we identified were true cases of PJI of the knee. Similar methods for identifying cases of PJI of the knee in the NIS have been previously reported [8, 14]. We then used ICD-9-CM procedure code 84.17 to identify 912 patients who underwent AKA. Patients were separated into two groups based on those who either had an AKA or those who did not. ICD-9-CM diagnosis codes were used to exclude patients who underwent AKA due to malignancy or trauma (see Appendix, Supplemental Digital Content, The incidence of AKA after TKA was determined by dividing 912 by the number of primary TKAs identified from the NIS between 2010 and 2014. Demographic characteristics of patients were abstracted from the database. The NIS records patient income quartiles based on the median income of their ZIP code. A description of these methods and income quartile cutoffs can be found online in the description of the NIS data elements [9].

The incidence of AKA after TKA in our study population was 912 of 31,995 patients. Patients who underwent AKA generally had a higher prevalence of comorbidities. The AKA group included more patients with diabetes (344 of 912 [38%] versus 9317/31,995 [29%]; p < 0.001) and anemia (298 of 912 [33%] versus 7647 of 31,995 [24%]; p < 0.001), as well as a higher proportion of patients with peripheral vascular disease (166 of 912 [19]% versus 1433 of 31,995 [5%]) compared with patients with PJI who did not undergo AKA (Table 1).

Table 1.:
Characteristics of patients with periprosthetic joint infection after TKA

Statistical Analysis

We used a Chi-square test to analyze categorical variables between the two groups. Multivariate logistic regression was performed to identify if certain patient demographic factors were associated with AKA after adjusting for comorbidities and hospital characteristics. Statistical significance was interpreted as a p value < 0.05 (two-tailed). All calculations were performed using IBM SPSS Statistics for Macintosh, Version 23.0 (IBM Corp, Armonk, NY, USA).


Compared with the wealthiest income quartile by ZIP code, patients in the lowest income quartile by ZIP code were more likely to have an AKA (odds ratio [OR], 1.58; 95% confidence interval [CI], 1.25–1.98; p < 0.001), and compared with patients with private insurance, patients with Medicare (OR, 1.94; 95% CI, 1.55–2.43; p < 0.001) and Medicaid (OR, 1.86; 95% CI, 1.37–2.53; p < 0.001) were at higher risk to have an AKA (Table 2). Patients in the 25th percentile (OR, 1.38; 95% CI, 1.09–1.73; p = 0.006) to the 75th percentile of income by ZIP code (OR, 1.27; 95% CI, 1.01–1.61; p = 0.041) of income also were at an increased risk of undergoing AKA compared with the highest income quartile by ZIP code. Compared with patients older that 80 years, patients younger than 50 years were less likely to have an AKA due to PJI (84 of 4861 [2%]; OR = 0.66; 95% CI, 0.46–0.96; p = 0.034), whereas patients aged 50 to 64 years (320 of 12,525 [3%]; OR = 0.92; 95% CI, 0.71–1.20; p = 0.622) or 65 to 80 years (393 of 12,571 [3%]; OR = 0.83; 95% CI, 0.66–1.04; p = 0.133) had similar risks of sustaining an AKA compared to older patients (reference, 115 of 2948 [4%]).

Table 2.:
Adjusted risk of above-knee amputation*

There were no differences with regard to risk of AKA for white patients (670 of 24,004 [3%]; OR = 0.99; 95% CI, 0.77–1.26; p = 0.936) or black patients (95 of 3178 [3%]; OR = 0.95; 95% CI, 0.69–1.30; p = 0.751) when compared with others (reference, 83 of 3159 [3%]). When compared with female patients, male patients did not have a greater risk of AKA (OR = 1.02; 95% CI, 0.88–1.29; p = 0.818).


Although rare, AKA is a severe complication of PJI after TKA. Previous studies have highlighted that certain factors such as older or younger age, black race, female sex, and Medicare insurance may be associated with an increased likelihood of a patient undergoing AKA due to peripheral vascular disease compared with other patients [4,21]. George et al. [8] found that with regard to PJI or aseptic TKA complications, black race was associated with an increased frequency of AKA, particularly for black males. Given the poor functional outcomes associated with AKA, it is essential to identify the patients at greatest risk of this complication to appropriately risk stratify patients who have undergone TKA with subsequent PJI of the knee [6, 16, 17, 19]. We found that patients with low socioeconomic status or Medicaid or Medicare insurance were more likely to undergo AKA after PJI of the knee.

The limitations of our study include several inherent weaknesses of working with the NIS database. The use of ICD-9-CM codes and the nonlongitudinal nature of the NIS prevent us from knowing with certainty that all patients who were identified as having an AKA indeed had the amputation as a result of a PJI. However, given that our calculated incidence of AKA after TKA is consistent with those from other studies [8, 16, 19], it is likely that our sample is representative of patients undergoing AKA in the context of PJI of the knee. Other large administratively coded databases would not have allowed us to more definitively identify AKA after TKA because they do not have followup data in the range of several years that would be needed to definitively identify such cases of AKA. Potential inaccuracies in coding represent another weakness of the NIS database; however, multiple studies have indicated that administratively coded databases representatively code patient comorbidities in total joint arthroplasty, as well as accurately identify PJI of the knee [1, 2, 15]. Therefore, the results of our study can be interpreted with confidence that coding inaccuracies have not appreciably affected our results.

While the NIS represents a 20% representative sample of inpatient hospital discharges in the United States, it is possible that the NIS might not be as representative for a rare procedure like AKA. However, a large national database such as the NIS was required to generate a sufficient sample size and we did not focus our analysis on national estimates of the prevalence of this procedure. Therefore, our results should be interpreted with some caution, as they might not be completely representative of all patients. Additionally, our method of defining socioeconomic status by median household income by patient ZIP code is not patient-specific. However, previously published studies have used this method and it remains the most robust surrogate for socioeconomic status given the constraints of the database [5, 13]. Despite these limitations, researchers frequently use the NIS in cases in which diseases or procedures are very uncommon to obtain sufficient sample sizes. In this case, given the extreme rarity of AKA after PJI of the knee, we believe that a large national database such as the NIS remains the most feasible way to investigate the epidemiology of this complication despite its inherent limitations.

Our findings that patients with Medicaid or Medicare insurance or low socioeconomic status were more likely to experience AKA is consistent with previous research which found that Medicare or Medicaid patients with peripheral vascular disease more commonly underwent AKA than patients with private insurance [21]. These findings corroborate the proposition that patient socioeconomic status may influence physician decisions. For example, it has been reported that patients with Medicaid undergoing primary total joint arthroplasty incurred greater hospital costs than patients with private insurance [18]. Carr et al. [3] found that patients with PJI who were treated with AKA incurred approximately USD 5000 less in hospital costs compared with patients managed with arthrodesis. As hospitals and physicians explore new payment models to improve patient care while minimizing costs, patient outcomes should remain at the forefront.

Our finding that there was no difference in the incidence of AKA when patients were stratified by race or sex contradicts previously published research [4, 8]. In an analysis of the NIS, George et al. [8] found that black race was associated with AKA after TKA. These conflicting results could potentially be explained by differing years of the NIS used by each study and slightly different inclusion criteria for patients who underwent AKA. However, the calculated incidence of AKA after TKA was similar for both studies (0.14% versus 0.17%). Notably, the previous study did not adjust for insurance type or median income in their analysis [8]. On the other hand, our results are consistent with work by Son et al. [20] who found no difference in the likelihood of AKA after TKA for different races. Given that a randomized controlled clinical trial is likely not feasible to test whether black race predicts AKA after PJI of the knee, physicians should still be cognizant that black patients could be at increased risk of AKA after PJI of the knee, especially for those patients with lower socioeconomic status. On a population level, future research should explore strategies for reducing the incidence of PJI, particularly in patients from low-income areas.

There are limited data on the number of conservative procedures patients undergo before AKA. A small study by Sierra et al. [19] found that 13 of 25 (52%) of patients who underwent AKA after unsuccessful TKA did not undergo revision TKA, whereas 8 of 25 (32%) had two or more revision procedures. In another small study of seven patients, the average number of procedures between TKA and AKA was 6.9 [12]. Future research should more clearly outline the number of patients who undergo limb-sparing procedures before AKA. While we did not find racial disparities in the use of AKA after PJI of the knee, multiple studies have found that black race is associated with an increased incidence of AKA due to peripheral vascular disease [4]. Future studies should examine outcomes in the subset of patients with peripheral vascular disease who experience PJI after TKA.

In patients undergoing PJI treatment after TKA, individuals with Medicare, Medicaid, or low socioeconomic status were more likely to undergo AKA. No racial or sex disparities were observed with regard to risk of undergoing AKA after PJI of the knee. Given that reimbursement rates differ between government and private insurance and that hospitals may be financially penalized for readmissions within a 30-day period, it is plausible that risk or perceived risk of readmission could subtly factor in a physician’s PJI treatment decision. There is evidence to suggest that patients with recurrent PJI of the knee who undergo more conservative limb-sparing procedures such as arthrodesis have higher rates postoperative infections, transfusions, and higher hospital charges [3]. Given that functional outcomes are poor after AKA, future research should explore physician attitudes and biases about patient followup and outcomes after PJI in patients with Medicare or Medicaid insurance, from low socioeconomic backgrounds, or with increased medical comorbidities, such as peripheral vascular disease.


1. Bozic KJ, Bashyal RK, Anthony SG, Chiu V, Shulman B, Rubash HE. Is administratively coded comorbidity and complication data in total joint arthroplasty valid? Clin Orthop Relat Res. 2013;471:201-205.
2. Bozic KJ, Chiu VW, Takemoto SK, Greenbaum JN, Smith TM, Jerabek SA, Berry DJ. The validity of using administrative claims data in total joint arthroplasty outcomes research. J Arthroplasty. 2010;25(Suppl 6):58-61.
3. Carr JB 2nd, Werner BC, Browne JA. Trends and outcomes in the treatment of failed septic total knee arthroplasty: comparing arthrodesis and above-knee amputation. J Arthroplasty. 2016;31:1574-1577.
4. Durazzo TS, Frencher S, Gusberg R. Influence of race on the management of lower extremity ischemia: Revascularization vs amputation. JAMA Surg. 2013;148:617-623.
5. Ehrenpreis ED, Zhou Y. Hospital costs, length of stay and prevalence of hip and knee arthroplasty in patients with inflammatory bowel disease. World J Gastroenterol. 2017;23:4752-4758.
6. Fedorka CJ, Chen AF, McGarry WM, Parvizi J, Klatt BA. Functional ability after above-the-knee amputation for infected total knee arthroplasty. Clin Orthop Relat Res. 2011;469:1024-1032.
7. George J, Navale SM, Nageeb EM, Curtis GL, Klika AK, Barsoum WK, Mont MA HC. Etiology of above-knee amputation in the United States: is periprosthetic joint infection an emerging cause? Clin Orthop Relat Res. 2018;476:1951-1960.
8. George J, Navale SM, Schiltz NK, Siccha M, Klika AK, Higuera CA. Racial disparities in above-knee amputations after TKA: a national database study. Clin Orthop Relat Res. 2017;475:1809-1815.
9. Healthcare Cost and Utilization Project. NIS Description of Data Elements. Available at: Accessed January 10, 2018.
10. Healthcare Cost and Utilization Project. NIS overview. Available at: Accessed January 10, 2018.
11. Isiklar ZU, Landon GC, Tullos HS. Amputation after failed total knee arthroplasty. Clin Orthop Relat Res. 1994;299:173-178.
12. Khanna V, Tushinski DM, Soever LJ, Vincent AD, Backstein DJ. Above knee amputation following total knee arthroplasty: When enough is enough. J Arthroplasty. 2015;30:658-662.
13. Kimball MM, Neal D, Waters MF, Hoh BL. Race and income disparity in ischemic stroke care: Nationwide inpatient sample database, 2002 to 2008. J Stroke Cerebrovasc Dis. 2014;23:17-24.
14. Kurtz SM, Lau E, Watson H, Schmier JK, Parvizi J. Economic burden of periprosthetic joint infection in the united states. J Arthroplasty. 2012;27(Suppl 8):61-65.e1.
15. Lau EC, Son MS, Mossad D, Toossi N, Johanson NA, Gonzalez MH, Meller MM. The validity of administrative BMI data in total joint arthroplasty. J Arthroplasty. 2015;30:1683-1687.
16. Mozella A de P, da Palma IM, de Souza AF, Gouget GO, de Araújo Barros Cobra HA. Amputation after failure or complication of total knee arthroplasty: prevalence, etiology and functional outcomes. Rev Bras Ortop (English Edition). 2013;48:406-41.
17. Pring DJ, Marks L, Angel JC. Mobility after amputation for failed knee replacement. J Bone Joint Surg Br. 1988;70:770-771.
18. Rozell JC, Courtney PM, Dattilo JR, Wu CH, Lee G-C. Should all patients be included in alternative payment models for primary total hip arthroplasty and total knee arthroplasty? J Arthroplasty. 2016;31:45-49.
19. Sierra RJ, Trousdale RT, Pagnano MW. Above-the-knee amputation after a total knee replacement: Prevalence, etiology, and functional outcome. Bone Joint Surg Am. 2003;85:1000-1004.
20. Son MS, Lau E, Parvizi J, Mont MA, Bozic KJ, Kurtz S. What are the frequency, associated factors, and mortality of amputation and arthrodesis after a failed infected TKA? Clin Orthop Relat Res. 2017;475:2905-2913.
21. Zayed M, Bech F, Hernandez-Boussard T. National review of factors influencing disparities and types of major lower extremity amputations. Ann Vasc Surg. 2014;28:1157-1165.

Supplemental Digital Content

© 2019 by the Association of Bone and Joint Surgeons