Analysis of Incidence and Risk Factors for Periprosthetic Fracture after Total Knee Arthroplasty in South Korea from 2010 to 2020 Based on National Registry Data

Abstract

Background

Periprosthetic fracture (PPF) is a troublesome complication as it utilizes substantial healthcare resources. Recent studies about the epidemiology of PPF after total knee arthroplasty (TKA) are still lacking, and there is limited national-level analysis focusing on the comorbid chronic conditions as risk factors of PPF. This study used national registry data from South Korea and aimed to investigate the epidemiology of PPF following TKA between 2010 and 2020 and identify which comorbidities contributed to the risk of PPF.

Methods

Using Health Insurance Review and Assessment (HIRA) service data in South Korea, the incidence of PPF after TKA between 2010 and 2020 was evaluated and stratified by age and sex. Medical comorbidities were evaluated as possible risk factors for PPF using Cox regression analysis.

Results

PPF occurred in 14,429 patients, accounting for 2.37% of total TKA patients. The prevalence of PPF by sex was 2.50% in women and 1.64% in men. The PPF rate was 2.82% in under 60 years, 2.25% in 60 to 69 years, 2.42% in 70 to 79 years, 2.29% in 80 to 89 years, and 2.12% in over 90 years. Among 17 analyzed comorbidities, 11 were found to be associated with PPF after TKA. Severe liver disease (hazard ratio [HR], 1.303), hemiplegia (HR, 1.244), and dementia (HR, 1.206) were the top 3 risk factors. Although osteoporosis, pulmonary disease, peptic ulcer, and diabetes showed relatively low HRs than these top 3 factors, the incidence rates were higher.

Conclusions

PPF occurred in 2.37% of TKA patients in South Korea from 2010 to 2020. PPF rate was higher in women. To prevent PPF after TKA, proper patient management and education should be emphasized, particularly in patients with severe liver disease, hemiplegia, and dementia.

Periprosthetic fracture (PPF) is a troublesome complication to both patients and healthcare providers as it requires readmission, reoperation, and additional costs and utilizes substantial healthcare resources.¹⁾ According to Reeves et al.,¹⁾ 90-day readmission rate after treatment of PPF was 20% and 90-day mortality was 2%–3%. As total knee arthroplasty (TKA) becomes more prevalent, the number of PPF is also increasing. From 2000 to 2008, the number of PPF increased by over fourfold in the United States.²⁾ In South Korea, the annual number of PPF patients has increased from 1,322 in 2010 to 2,636 in 2017, and the burden of PPF is expected to rise by 10 times during the next 10 years.³⁾ Given that the PPF is a significant burden, it is worth exploring the risk factors of PPF. Risk factors can be divided into surgical and nonsurgical factors. Surgical risk factors include revision TKA, previous osteotomy, previous patellar resurfacing, implant malalignment, instability, and femoral notching.⁴⁾ Nonsurgical or patient-related risk factors for PPF include age, inflammatory arthritis, chronic use of corticosteroids, and osteoporosis.⁵⁾ However, most of the previous studies suggesting these risk factors for PPF are based on the small patient groups or relatively outdated.⁶,⁷⁾ Recent studies about the epidemiology of PPF after TKA are still lacking, and there is limited national level analysis focusing on the comorbid chronic conditions as risk factors of PPF.

This study used national registry data from South Korea and aimed to investigate the epidemiology of PPF following TKA between 2010 and 2020 and identify which comorbidities contribute to the risk of PPF. The authors hypothesized that meaningful risk factors for PPF following TKA could be identified using national registry data.

METHODS

All data utilized in this study were collected through the Health Insurance Review and Assessment (HIRA) service in South Korea. Data collected between 2010 and 2020 were used. In South Korea, all citizens are required to be enrolled to the National Health Insurance Service, and every healthcare facility should submit data and information of medical procedures to HIRA for billing purposes. HIRA reviews the validity of theses submitted data and determines whether to reimburse the claimed expenses. The submitted data encompass various information including age, sex, primary diagnosis, surgical procedures, date of surgery, length of hospital stay, and conducted tests. Consequently, the data collected through HIRA effectively constitute a comprehensive record of all medical procedures performed in South Korea. As the collected data were de-identified by the HIRA, the requirement for Institutional Review Board approval and patient consent was waived for this study.

Data Collection

Initially, total primary TKA cases were screened using the procedural codes for TKA: N2072 and N2077. This resulted in 622,405 cases from 2010 to 2020. Among them, 13,800 cases had prior history of fracture before TKA, so were excluded. The following screening process to identify the PPF was conducted among the remaining 608,605 cases. There were 1,984 cases with diagnostic code for PPF (M966). In order to avoid any potential omissions, patients diagnosed with fractures around the knee joint were also identified: fracture of femur shaft, S723; fracture of distal femur, S724; fracture of proximal tibia, S821; fracture of tibia shaft, S822; and fracture of patella, S830. This resulted in 2,694 cases, 4,741 cases, 1,185 cases, 615 cases, and 4,960 cases, respectively. After removing the duplicate cases, 12,455 cases were identified as fracture around the knee. These cases were merged with the previously identified PPF cases, and after removing the duplicate cases, 14,429 cases were identified as PPF after TKA (Fig. 1). The PPF rate was defined as the ratio of total PPF patients to total TKA patients during the study period. Collected data were stratified by sex and age. Age group was divided into 5 groups: under 60 years; 60 to 69 years; 70 to 79 years; 80 to 89 years; and over 90 years.

Fig. 1. Flow diagram of patient selection. TKA: total knee arthroplasty.

To explore the association between PPF after TKA and medical comorbidities, Charlson comorbidity index was used and 14 diseases were selected for the analysis: acute myocardial infarction, peripheral vascular disease, cerebral vascular accident, dementia, pulmonary disease, connective tissue disorder, peptic ulcer, liver disease, diabetes, renal disease, cancer, metastatic cancer, severe liver disease, and hemiplegia. Additionally, we included osteoporosis, Parkinson disease, and Alzheimer disease, which were considered potential risk factors for PPF after TKA.⁸,⁹,¹⁰⁾ Using the International Classification of Disease (ICD) diagnostic codes for the 17 selected diseases, data were collected and used to analyze the association with PPF.

Statistical Analysis

All statistical analyses were conducted using SAS Enterprise Guide version 7.15 (SAS Institute Inc.) and R 3.5.1 (R Foundation for Statistical Computing). To identify the risk factors for PPF, a Cox regression model was employed, following the approach of previous studies on PPF risk.⁵,¹¹⁾ Cox regression models were utilized to determine the adjusted hazard ratio (HR) of 17 medical comorbidities as risk factors for PPF after TKA. The number of PPF and PPF rate were compared between sex and age groups. Statistical significance was determined at a 95% CI with a threshold of p-value < 0.05. The Kaplan-Meier curve was plotted to demonstrate the significance of the comorbidities for PPF.

RESULTS

Incidence of TKA and PPF

Among 608,605 cases of primary TKA included for the analysis, PPF occurred in 14,429 cases, resulting in 2.37% of PPF rate (Table 1). Age under 60 years showed the highest incidence of PPF at 2.82%. The PPF rates in individuals aged 60 to 69 years, 70 to 79 years, 80 to 89 years, and over 90 years were 2.25%, 2.42%, 2.29%, and 2.12%, respectively. When divided by sex, 12,876 PPF cases occurred in women, corresponding to a PPF rate of 2.50%, and 1,553 PPF cases occurred in men, resulting in a PPF rate of 1.64%.

Table 1. TKA and PPF Incidence Stratifed by Age and Sex.

Variable		TKA	Total PPF	PPF rate (%)
Total		608,605	14,429	2.37
Age (yr)
	< 60	33,932 (5.6)	957 (6.6)	2.82
	60–69	210,262 (34.6)	4,734 (32.8)	2.25
	70–79	303,414 (49.9)	7,343 (50.9)	2.42
	80–89	60,338 (9.9)	1,381 (9.6)	2.29
	> 90	659 (0.1)	14 (0.1)	2.12
Sex
	Male	94,443 (15.5)	1,553 (10.8)	1.64
	Female	514,172 (84.5)	12,876 (89.2)	2.50

Values are presented as number (%) unless otherwise indicated.

TKA: Total knee arthroplasty, PPF: periprosthetic fracture.

Comorbidities as Risk Factors of PPF

Among 17 analyzed comorbidities, 11 were found to be associated with the PPF after TKA (Table 2): severe liver disease, HR 1.303; hemiplegia, HR 1.244; dementia, HR 1.206; diabetes, HR 1.181; renal disease, HR 1.159; cerebral vascular accident, HR 1.148; osteoporosis, HR 1.102; liver disease, HR 1.100; connective tissue disorder, HR 1.086; peptic ulcer, HR 1.082; and pulmonary disease, HR 1.061. For the top 3 diseases with the highest HR, additional Kaplan-Meier curves were plotted (Figs. 2, 3, 4).

Table 2. Adjusted Hazard Ratio and p-value of Comorbidities.

Comorbidity	No PPF (n = 594,176)	PPF (n = 14,429)	Adjusted HR	95% CI	p-value
Severe liver disease	1,542 (0.3)	62 (0.4)	1.303	1.015–1.674	0.038
Hemiplegia	2,899 (0.5)	102 (0.7)	1.244	1.021–1.515	0.030
Dementia	27,760 (4.7)	806 (5.6)	1.206	1.118–1.301	< 0.001
Diabetes	211,229 (35.6)	5,663 (39.3)	1.181	1.141–1.221	< 0.001
Renal disease	19,260 (3.2)	485 (3.4)	1.159	1.057–1.270	0.002
Cerebral vascular accident	90,059 (15.2)	2,572 (17.8)	1.148	1.099–1.201	< 0.001
Osteoporosis	347,070 (58.4)	9,077 (62.9)	1.102	1.064–1.142	< 0.001
Liver disease	27,813 (4.7)	702 (4.9)	1.100	1.019–1.188	0.015
Connective tissue disorder	78,870 (13.3)	2,183 (15.1)	1.086	1.038–1.137	< 0.001
Peptic ulcer	226,657 (38.2)	6,230 (43.2)	1.082	1.047–1.119	< 0.001
Parkinson disease	7,666 (1.3)	222 (1.5)	1.081	0.946–1.236	0.254
Pulmonary disease	232,262 (39.1)	6,121 (42.4)	1.061	1.026–1.097	< 0.001
Acute myocardial infarction	34,287 (5.8)	926 (6.4)	1.032	0.965–1.104	0.356
Alzheimer disease	4,227 (0.7)	1,121 (0.8)	1.032	0.852–1.250	0.747
Peripheral vascular disease	50,950 (8.6)	1,235 (8.6)	1.012	0.953–1.074	0.702
Metastatic cancer	1,628 (0.3)	35 (0.2)	0.999	0.712–1.401	0.996
Cancer	33,952 (5.7)	723 (5.0)	0.966	0.895–1.043	0.374

Values are presented as number (%) unless otherwise indicated.

PPF: periprosthetic fracture, HR: hazard ratio.

Fig. 2. Kaplan-Meier graph depicting the incidence of periprosthetic fractures according to the presence of severe liver disease. The group with severe liver disease (blue dotted line) is compared to the group without severe liver disease (red solid line) in terms of periprosthetic fracture incidence. The blue shaded area represents 95% CI.

Fig. 3. Kaplan-Meier graph depicting the incidence of periprosthetic fractures according to the presence of hemiplegia. The group with hemiplegia (blue dotted line) is compared to the group without hemiplegia (red solid line) in terms of periprosthetic fracture incidence. The blue shaded area represents 95% CI.

Fig. 4. Kaplan-Meier graph depicting the incidence of periprosthetic fractures according to the presence of dementia. The group with dementia (blue dotted line) is compared to the group without dementia (red solid line) in terms of periprosthetic fracture incidence. The blue shaded area represents 95% CI.

DISCUSSION

PPF is a troublesome complication both to patients and surgeons. Considering the increasing life expectancy and the rise in comorbidities, the importance of PPF following TKA is growing, making it meaningful to examine this issue. The purpose of this study was to evaluate the incidence of PPF and analyze the comorbidities as risk factors of PPF after TKA. There were 2 primary findings in this study. First, the overall PPF rate was 2.37%, with a higher incidence in women. Although the rate did not differ substantially across age groups, it was highest in those under 60 years. Second, among 17 analyzed comorbidities, 11 were found to be associated with PPF after TKA, with severe liver disease, hemiplegia, and dementia being the top 3 factors.

The overall PPF rate after TKA in South Korea from 2010 to 2020 was 2.37%. It was comparable to the result from another national registry-based study conducted in South Korea between 2005 and 2018 by Ko et al.,¹²⁾ which reported 3.10% of PPF rate. When divided by sex, PPF rate was higher in women. There was conflicting evidence about sex as a risk factor of PPF. Meek et al.⁵⁾ used Scottish national database and showed women as a risk factor of PPF. Bengoa et al.⁴⁾ also claimed women as a risk factor of PPF. On the contrary, in another national database study in South Korea by Ko et al.,¹²⁾ men were the risk factor of PPF after TKA. Singh et al.¹¹⁾ used their clinic’s own registry data and reported that sex was not associated with PPF. Women are more susceptible to osteoporosis due to their hormonal nature, and the risks of osteoporotic fracture have been reported to be higher than those in men.¹³⁾ In this context, the higher PPF rate in women in our study seems reasonable, but the variation across the literatures suggests that there may be other contributing factors besides sex. Regarding age, there are conflicting studies on the influence of age on PPF risk. In our study, the PPF rate was highest in the age under 60 years, although it did not differ substantially across age groups. TKA was performed most frequently at ages of 60–69 years and 70–79 years, and the PPF incidence showed the same pattern. Pornrattanamaneewong et al.¹⁴⁾ reported that their PPF group was significantly older than the control group. In a study by Meek et al.,⁵⁾ female patients aged over 70 years were at risk of PPF after total hip and knee arthroplasty. On the contrary, in a study by Ko et al.,¹²⁾ PPF risk was increased in young patients. Singh et al.¹¹⁾ showed the risks of PPF according to age were in U-shape, meaning age under 60 years and over 80 years were more vulnerable to PPF. As previously mentioned, studies have reported differing results regarding the impact of age and sex on PPF risk. Each study was conducted based on different databases, suggesting that age and sex do not have a universal effect but rather exhibit varying influences depending on the population. Therefore, to minimize such confounding factors, a national-level big data analysis is necessary, which underscores the significance of this study. Ko et al.¹²⁾ used the same HIRA data as in our study, but reported an opposite result, identifying men as a risk factor for PPF. This discrepancy is likely due to their inclusion of revision TKA in the analysis. Given the difference in PPF incidence between revision and primary TKA, we believed it was meaningful to analyze them separately. Therefore, the current study focused only on primary TKA.

Eleven comorbidities were found to be associated with PPF after TKA. Identified risk factors were severe liver disease, hemiplegia, dementia, diabetes, renal disease, cerebral vascular accident, osteoporosis, liver disease, connective tissue disorder, peptic ulcer, and pulmonary disease. Among them, severe liver disease, hemiplegia, and dementia were found to be the most impactful factors. In a meta-analysis done by Liang et al.,¹⁵⁾ patients with liver cirrhosis (LC) had increased risk of fracture. Hepatic encephalopathy is known to be associated with an increased risk of fall, and this might be an explanation of LC being a risk factor of PPF.¹⁶⁾ Another possible explanation is the patient’s reduced activity level. Physical activity is reduced in patients with advanced liver disease.¹⁷⁾ Lim et al.¹⁸⁾ used patient-reported outcome measures (PROMs) to find out the risk factor of PPF after TKA and suggested that physical functioning and vitality subscales of Short Form-36 were significant predictors for PPF. This implies the importance of reduced patient’s activity level in the occurrence of PPF. In that regard, it is not surprising that hemiplegia and dementia were also found to be significant risk factors of PPF. These results are consistent with recent reports suggesting an association between the physical activity level and fracture risk.¹⁹⁾ Another interesting point of this study is that HR of osteoporosis was not as high as expected, given the general perception of osteoporosis as a high-risk factor of fracture. This suggests that a patient’s typical level of mobility may be a more important predictor of fracture risk than bone density alone. In addition, metabolic disease such as diabetes and renal disease showed higher HR. Bone metabolism is altered in CKD patients and this phenomenon is now called chronic kidney disease-mineral bone disorder (CKD-MBD).²⁰⁾ Patients with type 2 diabetes have higher risk of fracture for a given bone marrow density (BMD) than nondiabetic population.²¹⁾ As advanced liver disease, renal disease, and diabetes are multifactorial diseases and often accompany osteoporosis, HR of these conditions might have been higher than the HR of osteoporosis alone.²²,²³,²⁴⁾ However, although relatively low in HR, the incidences of osteoporosis, peptic ulcer, pulmonary disease, and diabetes were higher than those of severe liver disease, hemiplegia, and dementia. Considering the high prevalence, the appropriate management of these other comorbidities should not be neglected.

Not much studies have evaluated medical comorbidities as risk factors of PPF after TKA (Table 3). According to Singh et al.,²⁵⁾ peptic ulcer disease and chronic obstructive pulmonary disease (COPD) were associated with higher risk of postoperative PPF. Their study design was similar to our study in that they also used Charlson comorbidity index to identify the possible risk factors of PPF, but the revealed risk factors were different from ours. In their study, the number of included TKA was relatively small, and the data were originated from their own institution. In addition, they used data from 1989 to 2008. The difference of the results may be attributable to different nationality, time, and source of data. Bell et al.³⁰⁾ and Metikala et al.²⁷⁾ both evaluated LC as a risk factor of PPF and revealed its potential risk of developing PPF. Although it was comparable to our results and they also used national registry data, they focused on LC only. In another study by Pornrattanamaneewong et al.,¹⁴⁾ PPF group had more dyslipidemia and Parkinson disease, but multivariate analysis showed no statistical significance. However, the number of involved TKA was small. As for osteoporosis, Harris et al.²⁶⁾ and Kang et al.²⁸⁾ showed that low BMD was associated with PPF after TKA. Furthermore, Holzer et al.²⁹⁾ used a fracture risk assessment tool and suggested its usefulness in estimating individual PPF risk after TKA. Although there are several studies about medial risk factors of PPF, according to our knowledge, only limited research focused on the PPF occurrence after primary TKA based on the national registry data. Patients undergoing TKA often have many comorbidities, so overall care of the patient’s general health condition is necessary to prevent PPF.

Table 3. References of Medical Risk Factors for Periprosthetic Fracture after TKA.

Study	Study design	Country	Risk factors addressed in the study	Outcome
Harris et al. (2024)26)	418,054 TKA	US	Osteoporosis	Patients with osteoporosis had a nearly twofold increased risk of 5-year revision for PPF after TKA.
Metikala et al. (2023)27)	558,256 TKA (2,363 PPF)	US	LC	LC patients had a higher likelihood of developing PPF.
Kang et al. (2023)28)	5,364 TKA (24 PPF)	South Korea	BMD	Osteoporosis with BMD < –2.8 was associated with PPF during or 1 month postoperatively after TKA.
Holzer et al. (2023)29)	167 PPF (137 after THA, 30 after TKA)	Austria	FRAX	FRAX might be used to estimate individual PPF risk in patients following THA and TKA.
Bell et al. (2021)30)	1,734,568 TKA	US	LC	LC was associated with increased rates
Pornrattanamaneewong et al. (2021)14)	120 TKA (24 PPF, 96 controls)	Thailand	Deyo-Charlson comorbidity index Diabetes Hypertension Dyslipidemia Cardiovascular disease Thyroid disease Parkinson disease	PPF group had more dyslipidemia and Parkinson disease in univariate analysis, but multivariate analysis showed no statistical significance.
Singh et al. (2011)25)	17,433 TKA (188 PPF)	US	Cardiac disease Peripheral vascular disease Cerebrovascular disease moderate-severe renal disease Peptic ulcer disease COPD Diabetes Connective tissue disease Cancer Other (dementia, liver disease, AIDS)	The presence of peptic ulcer disease and COPD were associated with higher risk of postoperative PPF.

TKA: total knee arthroplasty, PPF: periprosthetic fracture, LC: liver cirrhosis, BMD: bone mineral density, THA: total hip arthroplasty, FRAX: fracture risk assessment tool, COPD: chronic obstructive pulmonary disease, AIDS: acquired immune deficiency syndrome.

There are several limitations to this study. Most of all, HIRA data did not provide detailed information. There was no available information about characteristics of each PPF case and the severity of the comorbidities. Therefore, an in-depth analysis of each case was not possible. In addition, in HIRA database, all periprosthetic fractures (hip, knee, shoulder, etc.) are labeled with the same diagnostic code (M966) and it does not differentiate right and left. Using this code to screen for PPF cases may have led to an overestimation of PPF occurrence after TKA. Therefore, we first screened patients who had undergone primary TKA and then identified PPF codes within that group. Through this process, we aimed to filter unnecessary data like PPF after hip and shoulder arthroplasty. However, another resulting limitation is that we could not accurately determine the annual incidence of PPF. As we selected cases of PPF occurring between 2010 and 2020 among patients who underwent TKA during this period, we were unable to include cases where TKA was performed before 2010 but PPF occurred between 2010 and 2020. Therefore, the PPF rate reported in this study was calculated based on cases occurring among TKA patients who underwent surgery between 2010 and 2020 and the rate for each individual year could not be determined. The study period was also relatively short. PPF risk would have been higher with a longer evaluation period. Although this study has limitations due to the constraints of the information provided by HIRA data, the use of national-level big data likely compensates for some of these limitations. Additionally, examining risk factors of PPF may provide further insights for managing patients after TKA. Considering that the top 3 factors contributing to PPF risk are severe liver disease, hemiplegia, and dementia, attention to patients’ mobility and frailty may be necessary. Although patients’ activity levels fall outside the scope of this study, further research on this topic is warranted.

In conclusion, PPF occurred in 2.37% of TKA patients in South Korea from 2010 to 2020. PPF rates were higher in women. To prevent PPF after TKA, proper patient management and education should be emphasized, particularly in patients with severe liver disease, hemiplegia, and dementia.