Abstract
Introduction:
Oxidised zirconium-on-polyethylene (OxZroP) total hip arthroplasty prostheses are increasing in popularity as an alternative to other bearing types. The surface characteristics and wear properties of OxZroP result in less metal and polyethylene wear which may be protective against microbiological seeding of the synovial joint space. The purpose of this study was to compare the rates of infection and inflammatory diagnoses between THAs using OxZroP heads with ceramic-on-polyethylene (CoP) and metal-on-polyethylene (MoP) designs.
Methods:
This study queried the Premier PINC AI Healthcare Database (PHD) (Premier Inc.) for all primary total hip arthroplasties from 2017 to 2022. Infection-related ICD10 codes were collected at 30 days, 3, 6, 12 months and 2 years postoperatively. Positive results were noted to be either PJI-related diagnosis codes, or the combination of an arthroplasty-related code with infection-related procedure codes. The rates of infection/inflammatory reaction were reported.
Results:
OxZroP THA bearings demonstrated a decreased rate of prosthetic joint infection or inflammatory reaction diagnoses, at up to 2 years post-surgery, compared to MoP bearings an effect that was maintained when cross-referenced with imaging procedure codes. When compared to CoP hips OxZroP demonstrated lower odds reduction of infection or inflammatory reaction with diagnosis codes at up to 6 months, however, showed equal odds to CoP at later timepoints.
Conclusions:
OxZroP bearings for primary total hip arthroplasty demonstrated a reduced odds ratio of prosthetic joint infection and inflammatory-related diagnoses than ceramic-on-polyethylene bearings before 6 months and maintained lower odds than metal-on-polyethylene bearings at up to 2 years. Further studies are needed to obtain longer-term follow-up and understand the mechanism for this observed reduction in prosthetic joint infection and inflammatory-related diagnoses.
Introduction
Total hip arthroplasty (THA) continues to be one of the most popular elective surgeries performed worldwide, and the number of cases is projected to continue to grow at a high rate over the next 40 years.1,2 As this volume increases it is important to focus on improving outcomes and longevity of total hip arthroplasty procedures, specifically when it comes to the most common complication of total hip replacement, prosthetic joint infection (PJI). 3 Despite significant advancements in the arthroplasty field, PJI rates have essentially stayed static over the last 2 decades. 4 While risk factors such as high body mass index (BMI), diabetes, and smoking status have been identified and targeted, PJI remains a major risk following THA and improvements continue to be needed. 5
2 of the more popular bearing surfaces for prostheses used in THA are metal-on-polyethylene (MoP), which uses a metal trunnion with a metal head articulating with a polyethylene acetabular liner, and ceramic-on-polyethylene (CoP) in which the head is made of ceramic material. CoP has gained favour over MoP due to the decreased risk of metal debris from trunnion wear in MoP which can cause elevated cobalt and chromium ion levels, adverse local soft tissue reaction (ALTR), osteolysis and destructive pseudotumour.6,7 Importantly, CoP implants have also demonstrated a significantly lower incidence of PJI than MoP implants. 8 This is thought to be due to the inflammatory environment from MoP hips leading to the recruitment of large numbers of leukocytes including phagocytic cells that may allow penetration of bacterial pathogens into the joint via a “trojan horse” mechanism.9,10 The mechanical wear and implant properties of CoP prostheses compared to MoP are thought to be what lead to the decrease in PJI risk due to the theoretical decreased infection via the inflammation pathway.
Components made with oxidised zirconium (OxZr) have emerged as alternatives to cobalt chromium or ceramic heads as the articulating surface with the polyethylene liner. This alloy has a low-friction surface that is composed of the superficial zone of the alloy itself rather than a surface coating and is resistant to abrasive wear.11,12 OxZr implants continue to grow in popularity due to their wear properties and excellent clinical performance. No studies to date have evaluated the incidence of PJI when comparing OxZr-on-polyethylene (OxZroP) components with MoP and CoP prosthesis in total hip arthroplasty.13–15 The primary purpose of this study was to compare the PJI rates between bearings using OxZroP and MoP as well as CoP prostheses.
Materials and methods
Data
This was a retrospective study of patients undergoing elective THA from 2017 to 2021 using the Premier PINC AI Healthcare Database (PHD) (Premier Inc.). The PHD is an extensive, U.S. hospital-based, service-level, all-payer database that contains information on inpatient discharges, primarily from geographically diverse non-profit, nongovernmental, and community and teaching hospitals and health systems from rural and urban areas. It does not include outpatient-designated surgeries. More than 135 million visits, with over 13 million per year since 2012, have been included in the PHD, representing approximately 25% of annual United States inpatient admissions. The PHD has been certified as deidentified, and all data complied with the Health Insurance Portability and Accountability Act (HIPAA). The study design was based on publicly available registry data and thus exempt from institutional review board (IRB) approval.
Patient selection
This study compared patients who underwent elective THA with either oxidised zirconium- (OXINIUM, Smith & Nephew, Memphis, TN, USA) on-polyethylene (OxZroP), COP, or MOP implants. Regarding polyethylene, data from the American Joint Replacement Registry (AJRR) notes that during the study period (2017–2022), approximately 96% of elective total hip arthroplasties have used highly crosslinked polyethylene. 16 Therefore, for purposes of the study the assumption that highly crosslinked polyethylene use was consistent across all groups was made, and a sub-analysis based on type of polyethylene was not deemed necessary. Patients were identified using ICD-10 procedural codes from July 2017 to June 2022. Further inclusion criteria included age ⩾21 years. Exclusion criteria included patients with simultaneous bilateral procedures, cases of hemiarthroplasty, resurfacing, revision procedures, cases of malignancy, and cases of THA performed for hip fracture. We also excluded patients with unmatched information between the billing records and procedure records, for instance, patients with discrepancies between the implant type reported in the billing records and ICD codes.
Study variables
Patient demographics including age, gender, race, cemented or cementless procedure and year of procedure were collected. We also considered hospital-related variables such as region, size (beds size), location (rural or urban), and hospital teaching status. Furthermore, we queried specific comorbidities including hypertension, obesity, congestive heart failure, diabetes mellitus, myocardial infarction, depression and Charlson Comorbidity Index (CCI). These were assessed based on diagnosis records between the index admission and 6 months before the index date.
The incidence of infection related to the arthroplasty procedure was the primary outcome assessed during the follow-up period from the index date to 6 months after the index date. ICD-10 diagnosis codes were used to identify diagnosis records of periprosthetic infection. The codes chosen were based on the combination used by the American Joint Replacement Registry (AJRR) to denote PJI and were listed as secondary diagnosis codes on the index admission or as primary or secondary diagnosis codes on any readmission or outpatient visit within the 30, 90, or 180 days after the index admission. These include codes for infection as well as inflammatory reaction and the selection of these ICD-10 codes was based on the recommendations of the AJRR. Inclusion of codes for inflammatory reaction has been validated by multiple studies and noted to be accurate in capturing the majority of PJI cases which are often missed if inflammatory reaction codes are excluded.17,18 To enhance the accuracy of identification of incidence of infection and inflammatory reaction, positive diagnosis codes were cross-referenced with procedure codes for concomitant hip x-ray, magnetic resonance imaging (MRI), or microbiology procedure such as arthrocentesis, blood culture, or arthrotomy 90 days before or after the infection and inflammatory reaction diagnosis date.
Statistical analysis
Patient demographics and clinical data were reported before and after propensity score matching (PSM). The chi-square test or Fisher exact test was used for categorical variables to test the distribution difference between OxZroP and CoP or MoP cohorts. The student’s t-test was used for continuous variables to test the difference between OxZroP and CoP or MoP cohorts.
PSM was performed to minimise confounding effects between the OxZroP and CoP or MoP cohorts. First, the multivariable logistic regression was used to calculate the propensity score for each patient in comparing groups with based on age, gender, year of procedure, cemented or cementless procedure, patients’ medical history during the baseline period, including hypertension, obesity, congestive heart failure, diabetes mellitus, myocardial infarction, depression, and CCI and hospital characteristics (region, size, location, and teaching status) as independent variables. Then, the nearest neighbor method with a caliper of 0.25 was used to perform the one-to-one matching between OXINIUM and CoP or MoP cohorts with the closest propensity score. The matching cohorts were further assessed for the balance of each covariate between cohorts, with a p-value > 0.05 or a standard mean difference (SMD) <0.1 considered well-balanced. Following PSM, 8262 were in each cohort of OXINIUM and MOP, and 8113 were in each cohort of OxZroP and COP. For the incidence of infection and inflammatory reactions after THA, the logistic regression model was used to estimate the percentage of event, odds ratio (OR), 95% confidence interval (CI) of OR, and p-value. SAS for Windows Version 9.4 (SAS Institute) was used for all statistical analyses. Statistical significance was noted for all analyses at p < 0.05.
Demographics
Following propensity score matching, there were no significant differences in demographics including age, reported race, sex, or use of cement between patients who received OxZroP or CoP implants (Table 1). There were also no observable significant differences in relevant medical comorbidities or CCI between the OxZroP patients and CoP patients (Table 2). When comparing hospital data, there were no differences in hospital size, hospital location, teaching status or region (Table 3).
Baseline demographic characteristics for propensity-matched patients in the oxidised zirconium-on-polyethylene (OxZroP), ceramic-on-polyethylene (CoP) and metal-on-polyethylene (MoP) groups.
SD, standard deviation.
Comorbidity-related demographic data for propensity-matched patients in the OxZroP, CoP and MoP groups.
OxZroP, oxidised zirconium-on-polyethylene; CoP, ceramic-on-polyethylene; MoP, metal-on-polyethylene; BMI, body mass index; SD, standard deviation.
Comparison of hospital data for propensity score matched patients in the OxZrop, CoP and MoP groups.
OxZroP, oxidised zirconium-on-polyethylene; CoP, ceramic-on-polyethylene; MoP, metal-on-polyethylene.
There were no demographic differences after one-to-one propensity score matching between patients who underwent THA with OxZroP implants and M0P implants (Table 1). Similarly, there were no differences in any medical comorbidities or between the hospital data for these 2 cohorts (Tables 2 and 3).
Results
Demographics
Following propensity score matching, there were no significant differences in demographics including age, reported race, sex, or use of cement between patients who received OxZroP or CoP implants (Table 1). There were also no observable significant differences in relevant medical comorbidities or CCI between the OxZroP patients and CoP patients (Table 2). When comparing hospital data, there were no differences in hospital size, hospital location, teaching status or region (Table 3).
There were no demographic differences after one-to-one propensity score matching between patients who underwent THA with OxZroP implants and MoP implants (Table 1). Similarly, there were no differences in any medical comorbidities or between the hospital data for these two cohorts (Tables 2 and 3).
A total of 388,656 inpatient primary total hip arthroplasties were identified. After excluding patients who underwent bilateral THA, resurfacing or revision procedures, patients under the age of 21, patients with missing demographic information, and patients who underwent THA for fracture we were left with 289,038 unique THA procedures. Patients with discrepancies between the type of implant reported in billing data and in operative reports were excluded, resulting in a final study cohort of 123,698 THA patients. CoP made up the highest number of THAs (112,294), while 32,484 were MoP and 8692 used OxZroP components. For later timepoints beyond 6 months, a total of 6912 OxZroP THAs met criteria for inclusion with adequate follow-up and were included for analysis.
Oxidised zirconium-on-polyethylene versu ceramic-on-polyethylene
OxZroP had significantly lower odds of a PJI-related infection or inflammatory diagnosis at 1 month (4.44% vs. 5.23%, p = 0.019), 3 months (5.34% vs. 6.14%, p = 0.0.028), and 6 months (6.06% vs. 6.88%, p = 0.035). There were no significant differences between OxZroP and COP at later timepoints including 1 year (6.15% vs. 6.06% p = 0.83), and 2 years (7.41% vs. 7.39% p = 0.97). OxZroP did demonstrate a lower rate of infections when diagnosis codes were combined with x-ray and then with a relevant procedure code for microbiological procedure (x-ray) than CoP (1.33% vs 1.99% p = 0.0027) overall. Another published method of validation using diagnosis codes combined with procedure codes for microbiologic procedures did not demonstrate any significance (p = 0.19) (Table 4).
Comparison of rates of medical record-diagnosed infection or inflammatory reaction with odds ratios and confidence intervals between OxZroP and CoP implants. Cross-referencing of diagnosis codes with validated procedure code data including imaging and microbiologic procedures is also included.
OxZroP, oxidised zirconium-on-polyethylene; CoP, ceramic-on-polyethylene; MoP, metal-on-polyethylene.
Note: values in bold indicate statistical significance.
Oxidised zirconium-on-polyethylene versus metal-on-polyethylene
OxZroP had a lower rate of infection or inflammatory reaction diagnosis than MOP at all observed timepoints, including at 1 month (4.45% vs. 5.42%, p = 0.0041), 3 months (5.36% vs 6.37%, p = 0.0061), 6 months (6.11 vs. 7.08%, p = 0.0122), 1 year (6.15% vs. 7.25%, p = 0.011), and 2 years (7.37% vs. 8.37%, p = 0.032). OxZroP further showed significance when diagnosis codes were combined with relevant imaging diagnoses (1.29% vs. 1.77%, p = 0.025). No significant difference was reported when diagnosis codes were paired with a relevant procedure code for microbiological procedure (0.45 vs. 0.53%, p = 0.46) (Table 5).
Comparison of rates of medical record-diagnosed infection or inflammatory reaction with odds ratios and confidence intervals between OxZroP and MoP implants. Cross-referencing of diagnosis codes with validated procedure code data including imaging and microbiologic procedures is also included.
OxZroP, oxidised zirconium-on-polyethylene; CoP, ceramic-on-polyethylene; MoP, metal-on-polyethylene.
Note: values in bold indicate statistical significance.
Discussion
This retrospective analysis found that the use of OxZroP was associated with a lower rate of early PJI-related diagnoses compared to CoP and lower rates than MoP at up to 2-year follow-up. Our methodology used one-to-one propensity matching of 8692 patients with adequate follow-up at up to 6 months, and a subset of 6912 patients with adequate follow-up extending to 2 years for the analysis. This study also looked at previously validated methods for confirming the diagnosis of PJI in large databases studies.17,18 1 such methodology, the combination of a relevant diagnosis code with procedure code for hip imaging demonstrated significance in comparisons of OxZroP to both CoP and MoP hips. The other method of combining a diagnosis code with a procedure code for either blood culture or guided hip aspiration, did not yield significance in either comparison arm. This is the largest study to date comparing OxZroP to CoP and MoP THA bearings. The strength of the study is the large sample size based on a large database from various US based data sets, which improves the generalisability of the findings. This result is statistically significant even after propensity score matching both the patient and treatment centre demographics.
In a recent study of MoP versus CoP THA bearings, Chisari et al. 8 found that MoP had significantly higher risk of PJI with 1 year minimum follow-up. The rate of PJI in CoP was 1.64% and the rate of adverse local tissue reaction was 0.87%. This suggests the risk of PJI was lower in CoP than MoP THA, which was attributable to a variety of factors including the material’s pro-inflammatory changes on soft tissue. Adverse local tissue reactions (ALTR) initially described with metal-on-metal (MoM) THA has also been described in MOP and is thought to be related to particulate from wear and trunnionosis. 19 While a direct link between ALTR and PJI has not been established, there have been cases of concomitant ALTR with PJI. 20 In a comparison of CoP to MoP THA, Tan et al. 21 showed CoP to have less corrosion and fretting compared to MoP. The worse wear profile of MoP THA may explain the higher PJI rates seen. On the other hand, Hexter et al. 22 showed no difference between PJI rates between MoP and CoP THA in a systemic review. Similarly, Ong et al. 23 and Parsons et al. 24 saw no difference in infection in comparisons between MoP and CoP.
Direct data comparison of OxZroP versus CoP THA has been sparse. Morison et al. 25 reported a PJI rate of 1 out of 46 (2.2%) in THA with OxZroP on cross-linked polyethylene. In a systematic review, the average rate of PJI for CoP was 0.38% and ranged from 0% to 2.3%. 22 Chisari et al. 8 reported the sum of the rate of PJI and ALTR was 2.51% in CoP, which is lower than the rates reported in CoP THA in the current study. This is likely attributable to difference in diagnosis capture between the studies. 1 reason the rate of infection and inflammation is higher in the current study is because a larger sample size would theoretically detect complication that occur at a relatively low rate, such as PJI, better than a smaller sample size. Notably, only 10,602 patients were in the prior study, compared to 123,698 in the current study. Interestingly, it has been published that ceramic-on-ceramic (CoC) bearings have a lower association with infection than CoP bearings which makes sense in the context of a larger discussion of material properties and wear-debris. 26 Indirect comparison can be made by extrapolating data from other large, registry studies with a similar design to the present study. Pitto and Sedel 27 queried the New Zealand Joint Registry over a 15-year period comparing rates of revision for deep infection between CoC, CoP, MoP and MoM hips. They noted no difference in infection risk at the early time point (6 months) between any groups including CoP and MoP, but over the duration of the entire study period CoC hips demonstrated significantly lower revision for deep infection than CoP, MoP and MoM hips. 27 This further suggests the need for future studies of a similar design comparing OxZroP directly to COC bearings.
Similar to comparisons with CoP THA, data directly comparing OxZroP and MoP THA has been lacking. The lower rates of PJI in OxZroP THA compared to MoP in the current study suggest it may behave differently from MoP THA. Hexter et al. 22 conducted a systemic review of bearing surfaces and PJI, and they reported a PJI rate of 0.85% in MoP, ranging from 0.73% to 3.7%. Morison et al. 25 conducted a randomised clinical trial assigning patients to MoP and OxZroP bearings, and they found PJI rate of 2.2% in OxZroP versus 4.4% in MoP. The current study suggests a possible difference in wear profile of OxZroP and MoP THA to explain the higher ATLR and PJI rate. The inside of the taper of OxZr heads is also ceramicised, thus likely making them behave more akin to ceramic heads with a lower propensity to cause metal-assisted crevice corrosion (MACC) and subsequent inflammation, as noted by Cartner et al. 28 in an implant retrieval study. This is consistent with Morison et al.’s results who noted no difference in wear rate of OxZroP compared to CoP at 6.8 years. 25 However, some literature does show in certain circumstances that OxZroP bearings may undergo corrosion comparable to metal bearings, although this may be more related to use of cobalt chromium stems in that particular study. 21
Theories have been suggested as to why certain bearing types may lead to higher infection rates. Pro-inflammatory changes from the bearing and any wear particles may cause leukocyte recruitment. Zhu et al. 10 conducted a study where methicillin-resistant Staphyloccus aureus (MRSA) and MRSA-carrying neutrophils were injected intravenously into a rat model of PJI, and they found rats injected with MRSA-carrying neutrophils to have higher rates of PJI. They suggested that PJI development is related to the release of bacteria from neutrophils, which may act as a “Trojan Horse.” Metal particles have been postulated to be more immunogenic than ceramic polyethene particles. Ceramic and polymeric debris activate the innate immune system and nonspecific reaction. On the other hand, metallic ion particulates stimulate both the innate and adaptive immune system to cause a type IV cell mediated hypersensitivity. 29 Good et al. 11 showed OxZroP to have 45% less wear compared to cobalt chromium (CoCr) heads when smooth, and 61% less than CoCr when roughened. This may explain why OxZroP has less particulate driven inflammation and subsequent infection driven by the “Trojan Horse” neutrophils.
Our study used robust methods to identify the incidence of PJI and inflammatory related diagnoses from this large database study. This methodology has been validated in PJI by Weinstein et al., 18 however, there are likely still some missed PJI diagnoses. Not all PJIs undergo imaging with hip x-rays or MRI beforehand and often the surgeon may forego a microbiologic procedure in the presence of an otherwise high index of suspicion based on clinical findings. 30 Several limitations exist in our study. As a large database study, the findings are generalisable, however it relies on proper coding of diagnosis and procedure codes at all institutions, which may not always be equally accurate. The indication and selection process for each type of head is unclear, raising the possibility of confounding bias and Type I error. Additionally, there are other confounding factors that may play a role in the development of PJI that are important considerations, such as surgical approach, use of perioperative antibiotics, transfusion rates or other patient-specific factors.
Conclusion
In this large database study, the use of OxZroP bearings demonstrated lower rates of PJI-related diagnosis codes than COP bearings at up to 6 months post-surgery. OxZroP similarly had lower rates of PJI-related diagnosis codes than MOP bearings at all timepoints up to 2 years follow-up. Further long-term studies comparing the infection and inflammatory reaction rates of OxZroP and CoP are warranted.
Footnotes
Declaration of conflicting interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: RS: Consultant: Smith & Nephew, Intellijoint.
TS: Consultant: Smith & Nephew.
CC and LN: Paid employees of Smith & Nephew.
All other authors declare that there is no conflict of interest.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The study was funded by Smith & Nephew Orthopaedics.
