May the 4th be with you: mixed-methods best-evidence synthesis on 4th-generation alumina–zirconia ceramic bearings in total hip arthroplasty

Abstract

Purpose

• To assess utility, benefits, and risks of 4th-generation alumina–zirconia ceramic pairings in elective total hip arthroplasty (THA).

Methods

• A comprehensive mixed-methods best-evidence synthesis using data from systematic reviews, randomized controlled trials (RCTs), prospective and retrospective cohort studies, as well as joint replacement registries, was conducted to estimate overall revision and survival rates, periprosthetic infection, bearing fractures, and noise phenomena with 4th-generation alumina–zirconia ceramic versus other tribological couplings in elective THA. The systematic review part across multiple databases was registered with PROSPERO (CRD42023418076), and individual study data were extracted for statistical re-analysis.

Results

• Twenty overlapping systematic reviews, 7, 17, and 8 references from RCTs, cohort studies, and joint replacement registries form the basis of this work. According to current best evidence, it is (i) 15–33 times more likely that 4th-generation alumina–zirconia pairings avoid a revision for infection than causing a revision for audible noise, (ii) 38–85 times more likely that 4th-generation alumina–zirconia pairings avoid a revision for infection than causing a revision for ceramic head fractures, and (iii) three to six times more likely that 4th-generation alumina–zirconia pairings avoid a revision for infection than cause a revision for ceramic liner fractures.

Conclusion

• Fourth-generation alumina–zirconia pairings in THA show a favorable benefit–risk ratio, with rare compound-specific adverse events and complications significantly outbalanced by long-term advantages, such as a markedly lower incidence of revision for infection.

Introduction

Charnley’s artificial hip joint concept (1, 2, 3, 4) represented a disruptive technology, which, like solid organ transplantation, fundamentally changed healthcare without proof by large-scale, multicenter randomized controlled trials (RCTs). Key principles of total hip arthroplasty (THA) remained virtually unchanged for more than 60 years. Current hardware is likely to survive 25 years and longer, exceeding the lifespan of most patients (5).

The durability and longevity of implants and patient outcomes depend on multiple intrinsic and extrinsic, both unalterable and modifiable variables, and their biological interactions (i.e. demography, comorbidity, medication, surgical approach, anatomical component positioning, soft-tissue balancing, restoration of leg length and axis, navigation, cemented or cementless fixation of the acetabular shell and femoral stem, platforms by different manufacturers, and many others). The tribologically optimal combination of ball and socket material remains a decisive factor for reducing wear and maximizing the likelihood of uneventful long-term survival (6).

To date, hard-on-soft (i.e. cobalt chromium molybdenum (CoCrMo) metal and alumina- or alumina–zirconia-ceramics-on-highly cross-linked polyethylene with or without antioxidants, HXLPE) and hard-on-hard (i.e. ceramic-on-ceramic) pairings are the most common in use, all of which have distinct risk–benefit profiles (7, 8).

The idea of using alumina ceramics as a fabric for THA dates back to the early 1970s and the French orthopedic surgeon pioneer Pierre Boutin (9). It needed, however, advanced manufacturing and processing technology, and another 25 years before ultrapure alumina oxide (e.g. BIOLOX forte, CeramTec, Plochingen, Germany) was certified for clinical use. Endeavors to strengthen the material even further, smoothen articulation, and minimize wear debris led to alumina–zirconia as the current industry standard, often called 4th-generation ceramics. A well-known proxy of this product line is BIOLOX delta (CeramTec), approved by authorities like the Food and Drug Administration (FDA) in 2003. Apart from clinical and healthcare issues, one must not neglect that, together with any other THA hardware and bearing, ceramics for medical purposes play a significant role in a highly competitive global market. Both BIOLOX forte and delta widely stand as a synonym of the alumina oxide and alumina–zirconia family and may represent the most frequently implanted brands worldwide, although market share data are difficult to retrieve and not publicly available.

Substantial international differences exist in the preferred choice of THA couplings (10). According to the most recent 2022 German Arthroplasty Registry (EPRD) report, ceramic-on-HXLPE accounts for 70% of all THAs in Germany, followed by ceramic-on-ceramic (8%) and metal-on-HXLPE (5%) (11). In Sweden, more than 70% and 15% of THAs are performed with metal-on-HXLPE and ceramic-on-HXLPE, respectively (12). Of the 89 000 primary THAs conducted across the UK in 2021, around 42 000 employed metal-on-HXLPE and ceramic-on-HXLPE pairings each (13). In the USA, 63% and 13% of all patients undergoing primary THA received ceramic-on-HXLPE and metal-on-HXLPE, respectively (14). In other words, the prior probability of a patient scheduled for THA to receive a certain bearing may, apart from individual health and demographic criteria, vary by 60% depending on nationality or citizenship. This diversity cannot be explained by scientific evidence alone. Implant availability or permit by postcode, insurance default, or provider preference without considering the current body of scientific information is dangerous for patients and the community and is likely to increase secondary tangible costs and complications in the long run.

Evidence-based decision-making should preferably rely on experimental data. Yet, despite the high frequency of THA (about two million procedures performed annually in the OECD countries alone (15)), less than 35 000 patients were enrolled in RCTs between 2000 and 2020 to study new developments or modifications like cementless fixation, short stems, novel bearings, etc., in a causal fashion (16, 17).

In such a situation, it is inevitable to consider real-world evidence as well (e.g. joint replacement registries and administrative and routine observational data) to provide an unbiased picture of the benefits, disadvantages, and risks of certain health technologies.

We herein employed best-evidence synthesis as a pragmatic and valid tool to compile experimental, quasi-experimental, and observational data (18, 19) to inform clinicians and healthcare authorities about the effectiveness and safety of 4th-generation alumina–zirconia couplings in THA.

We posed the patient and/or problem, intervention, control, outcome, time (PICOT) question, whether THA using 4th-generation (i.e. alumina–zirconia) ceramic head and/or liner pairings is associated with better outcomes compared to metal heads and/or any sort of polyethylene liners. This was operationalized as follows: (i) lower overall revision and/or higher component survival rates, (ii) lower revision rates for infection, and (iii) improved patient-reported function and/or quality of life, at any reported time of follow-up. We also assessed the incidence of adverse events and reactions typically reported with ceramic hips, such as squeaking and fractures, and how they should be traded off against their possible benefits.

Methods

Literature search

The systematic review portion of this work was registered with PROSPERO (CRD42023418076). Literature retrieval across multiple databases adhered to recommendations of the Cochrane Collaboration (20) and PRISMA guidelines (21). We employed PubMed and OVID Medline, OVID Embase, CINAHL (via EBSCO), and the Cochrane Library (Database of Systematic Reviews and Central Register of Controlled Trials) as premier sources of scientific information. This was followed by a hierarchical snowball procedure to ensure completeness of evidence and unbiased results. This process involved repeated search and extraction steps in the following order:

1. Primary identification of systematic reviews and meta-analyses.

2. Tabulation of studies included in systematic reviews to illustrate the intersecting set of references and overlapping evidence.

3. Electronic search for novel individual trials and studies hitherto not included in published systematic reviews.

4. Exploration of reference lists of all publications for relevant articles missed by the mentioned strategy.

5. Screening of trial registries, i.e. clinicaltrials.gov and current controlled trials.

6. Survey of major joint replacement registries like:

   1. Australia (https://aoanjrr.sahmri.com)

   2. Denmark (https://danskhoftealloplastikregister.dk)

   3. Germany (https://www.eprd.de/en)

   4. The Netherlands (https://www.lroi-report.nl)

   5. New Zealand (https://www.nzoa.org.nz/nzoa-joint-registry)

   6. Norway (https://www.kvalitetsregistre.no/register/muskel-og-skjelett/nasjonalt-register-leddproteser)

   7. Sweden (https://sar.registercentrum.se)

   8. UK (https://www.njrcentre.org.uk)

   9. USA (https://www.aaos.org/registries/registry-program/american-joint-replacement-registry).

All references were imported and stored in full-text PDF format in EndNote 20.5 (Clarivate Analytics LLC, USA).

Eligibility

We extracted data from systematic reviews and meta-analyses, individual RCTs, observational clinical studies, and total joint replacement registry reports comparing common ceramic couplings (i.e. ceramic-on-HXLPE (CoHXLPE) or ceramic-on-ceramic (CoC)) with alternatives like metal-on-HXLPE (MoHXLPE) in primary elective unilateral or bilateral THA for osteoarthritis.

Although we placed a focus on 4th-generation ceramics (i.e. alumina–zirconia compounds) approved in 2003, our search algorithm went back to January 1, 1994 (marking the introduction of alumina oxide BIOLOX forte ceramic heads (CeramTec, Plochingen, Germany)). This guaranteed high sensitivity and reduced the risk of missing clinical investigations on alumina–zirconia prior to approval by the FDA, EMEA, and other regulatory bodies. We included studies published until May 1, 2023.

We decided to exclude research incorporating metal-on-metal (MoM) pairings unless relevant information on CoC controls was provided, as there is substantial evidence MoM hip resurfacing and stemmed THA with large heads are no longer a viable treatment option because of excess revision rates, metallosis, and other complications (22, 23, 24).

Alumina–zirconia pairings from manufacturers other than CeramTec (e.g. Zimmer, Smith & Nephew, DePuy, Stryker, Ceraver Osteal, Saint-Gobain Desmarquet, Wright Medical, Kyocera, and many others) were reviewed and included as well. However, we had a high prior probability that it was more likely to identify and enroll clinical trials and observational data on BIOLOX delta than on its alumina–zirconia competitors.

We included studies enrolling male, female, and diverse populations ≥18 years of age, published in Danish, Dutch, English, French, German, Spanish, and Swedish, given the foreign language competence of reviewers. Using authors’ lines, affiliations, and trial registries, we investigated whether a certain trial or cohort was unique or just a long-term follow-up report of a previous one.

Enrollment criteria

To be eligible, studies had to provide

1. ≥1 unequivocal, clearly defined THA coupling including alumina–zirconia ceramic heads on alumina–zirconia or any polyethylene liner material,

2. ≥1 unambiguous, clearly operationalized categorical endpoint (e.g. (cumulative) surgical revision for any cause).

Exclusion criteria

As we attempted to provide a pragmatic, ubiquitously applicable risk–benefit estimate of 4th-generation alumina–zirconia components, we made no restrictions to (i) prosthetic metal hardware produced by any manufacturer, (ii) cemented or uncemented fixation, (iii) surgical access routes, (iv) additional procedures such as accelerated or fast-track rehabilitation, navigation, etc. However, we considered these variables for later sensitivity or subgroup analyses.

We excluded revision procedures, as well as THA performed for hip fractures. We also excluded laboratory, biomechanical, and animal experiments (including finite element models, clinical studies of surrogates like radiostereometry, Einzelbild-Röntgenanalyse (EBRA) etc.), and case series.

Data extraction and summary

Data from eligible references were compiled in a standardized electronic chart and sketched in a short profile. We abstracted the following:

1. Key methodological criteria (e.g. study type, country and study site(s), publication year, duration of recruitment, trial registration, a priori defined hypotheses and effect sizes, pre-defined statistical methods).

2. Trial, study, and/or individual patient baseline characteristics (demographics, co-morbidity, disease-specific variables, etc.), as provided by full-text publications and their online Supplements and appendices.

3. Treatment details (acetabular and femoral components, fixation, surgical access routes, adjunct procedures, etc.).

4. Categorical and continuous outcomes.

5. Duration of follow-up, sample sizes, numerators and denominators of binary outcomes and adverse events, means, medians, s.d ., interquartile ranges (IQR), ranges, and 95% CI for later data conversion and processing.

If main results were published as figures only (which often occurred with, but was not limited to survival curves), we extracted data by copying and reproducing the figure manually (including measures of distribution) based on jpgs embedded in Microsoft Excel sheets. Given our previous experience, we deliberated and refrained from automated tools like Plot Digitizer (25).

Exposure and outcome criteria

Key exposures

1. Different ball-and-socket pairings using alumina–zirconia and its comparators (i.e. CoHXLPE, CoC, MoHXPLE, and others).

2. Implant- and implantation-associated variables (i.e. cemented or cementless fixation, femoral head size, manufacturer, surgical approach, etc.).

Primary outcome

The primary outcome of this work was overall revision-free survival, expressed as median survival, proportion of unrevised THA at a certain time of follow-up, or hazard ratio (HR).

Co-primary outcomes

1. Incidence of and/or surgical revision for prosthetic infection, operationalized as deep surgical site infection (SSI) according to criteria of the Centers for Disease Control and Prevention (https://www.cdc.gov/nhsn/pdfs/pscmanual/9pscssicurrent.pdf), and definitions of the Philadelphia International Consensus Meeting (https://www.efort.org/wp-content/uploads/2013/10/Philadelphia_Consensus.pdf).

2. Incidence of and/or surgical revision for ceramic fractures (heads and/or liners), verified by advanced imaging (e.g. CT and/or MRI), intra-operative findings, or subsequent implant retrieval analysis.

3. Incidence of and/or surgical revision for squeaking and/or other noise (26, 27), reported by patients and/or verified by healthcare professionals during clinical examination.

Secondary outcomes

• Patient-reported outcome measures (PROM):

  1. Disease-specific instruments like Harris Hip Score (HHS) (28), Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) (29), and others.

  2. Generic health outcome questionnaires like the Short-Form 36 (SF-36) (30), Euro-QoL 5D (EQ-5D) (31) with their 3L and 5L scoring approaches (32), etc.

Rating of methodological quality

With regard to systematic reviews, we accepted the original authors’ qualitative rating of included studies once they adhered to validated and accepted rating schemes like the revised Cochrane Risk-of-Bias tool (RoB 2) (33), the Cochrane Risk Of Bias In Non-randomized Studies (ROBINS-I) instrument (34), the Grading of Recommendations Assessment, Development and Evaluation (GRADE) (35), and others. There are currently no unified or commonly accepted quality appraisal checklists or reporting guidelines for total joint arthroplasty registries, although some recommendations had been provided by the EMA and the CDC for other fields of interest (https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-registry-based-studies_en.pdf-0Individual studies, https://www.cdc.gov/datainteroperability/publichealthoptions/PH-registries-reporting.html). Studies not included in systematic reviews and identified by our advanced search strategy were methodologically assessed by the RoB 2 or ROBINS-I instrument. We employed the web-based robvis visualization tool (https://www.riskofbias.info/welcome/robvis-visualization-tool) to generate traffic-light plots, which were manually transferred to processable figures.

Data standardization and aggregation

Whenever possible, we derived raw numbers for all exposure and outcome variables to re-calculate or calculate consistent effect measures, i.e. risks and risk differences, means, mean and standardized mean differences, odds ratios (ORs), and risk ratios (RRs) with 95% CIs. This demanded some data transformation and approximations outlined in the Cochrane Handbook (36).

Typical narrowing formulas were:

1. Medians were set equal to means if no other information was available.

2. The S.D. was estimated as the range of data divided by 4.

3. Standard errors (SE) of proportions (p), given the sample size n were calculated as assuming a normal distribution, with upper and lower 97.5% confidence limits computed as p ± 1.96 · SE.

4. The upper 97.5% confidence limit in the case of null events was estimated at 3/n according to Hanley’s rule (37).

Where suitable, we aimed at quantitative data synthesis using meta-analytical methods. This included:

1. Random-effects meta-analysis of proportions.

2. Random-effects meta-analysis of OR, RR, and HRs.

3. Random-effects meta-analysis of (standardized) mean differences.

4. Meta-regression.

Individual and aggregated indices of effectiveness were reported with 95% CI. STATA MP 16.0 (StataCorp LLC) was employed for all quantitative analyses.

Results

Literature search

The full search strategy and number of identified studies in Medline, Embase, CINAHL, and the Cochrane Central Register of Controlled Trials (updated on May 21, 2023) are shown in Supplementary tables (see section on supplementary materials given at the end of this article). A PRISMA flowchart is depicted in Fig. 1.

Figure 1.

PRIMA flowchart.

We identified 20 systematic reviews and meta-analyses published between 2014 and 2023. Supplementary tables show the distribution and overlap of individual studies included in previous systematic reviews and meta-analyses. Together with additional searches, we revealed 57 citations from RCTs, 29 from prospective cohort studies, 100 from retrospective cohort studies, and eight reports from total joint arthroplasty registries. Forty-seven and 29 citations specifically referred to BIOLOX delta pairings, respectively, 17 of which reported results from RCTs.

Seven, 17, and eight references from RCTs, prospective, and retrospective cohort studies, along with data from eight joint replacement registries, form the basis of this work.

Overall revision-free survival and/or cumulative revision rates

A semi-quantitative profile of all studies enrolled in this investigation, together with their RoB 2 or ROBINS-I ratings, can be found in the Supplemental appendix. In general, RCTs (mainly FDA IDE trials) had a low to moderate risk of bias, whereas 12/17 cohort studies suffered from a serious risk of bias in ≥4 of seven domains.

A summary of overall revision-free survival rates with alumina–zirconia couplings observed in four RCTs (38, 39, 40, 41) and 14 cohort studies (42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55) is shown in Table 1.

Table 1.

Summary of overall component survival reported by RCTs and cohort studies.

Design	k	n	F/U, years		Overall survival, %
			Mean (s.d.)	Median, range	Mean (s.d.)	Median, range
RCT	4	368	5.8 (0.8)	5.6 (5.0–6.8)	95.4 (1.6)	94.9 (94.1–97.7)
Cohort studies	14	3912	6.3 (2.8)	5.9 (2.7–13.1)	98.3 (1.5)	98.5 (94.2–100.0)

Major hip joint registries typically documented outcomes of metal hardware (i.e. stems and sockets) from various manufacturers, as well as cemented and cementless fixation. Classification of couplings, however, was limited to general principles (i.e. CoC, CoP, MoP, and others), without striving for more detailed analyses (e.g. alumina oxide versus alumina–zirconia).

In general, registry data favored the overall component survival of CoHXLPE over MoHXLPE bearings. Adjusted estimates also favored CoC over MoHXLPE, although some results remained conflicting (Fig. 2).

Figure 2.

Graphical summary of registry data shown in Table 2.

The UK’s NJR (56) allows for in-depth analyses of bearings on request, thereby bridging registries, systematic reviews, RCTs, and individual studies. The sponsor of this investigation commissioned the NJR to estimate survival rates and complications with BIOLOX delta CoC and CoP versus MoP bearings (Table 2).

Table 2.

Key results of the commissioned bivariate in-depth analyses of the UK’s NJR.

Variable	In-depth analysis on CoC		In-depth analysis on CoP
Pairing	CoC	MoP	CoP	MoP
Patients	112 194	648 988	222 951	648 988
THA	128 092	739 808	246 757	739 808
Mean age	59	74	64	74
Median BMI	28	28	28	28
Male patients	47%	36%	43%	36%
ASA
I	26%	10%	17%	10%
II	66%	69%	70%	69%
≥III	8%	21%	13%	21%
Reason for revision, revised (O/E)
Unexplained pain	408 (0.96)	1198 (1.03)	193 (0.57)	1198 (1.15)
Infection	504 (0.88)	3184 (1.04)	821 (0.90)	3184 (1.05)
Socket fracture	205 (1.71)	62 (0.44)	20 (1.12)	62 (1.00)
Head fracture	27 (1.55)	14 (0.62)	16 (1.70)	14 (0.72)

Estimates with CoP differ because of different multivariate model structures.

O / E, observed to expected ratio.

Cumulative revision rates for MoP, CoC, and CoP were 5.22%, 4.34%, and 3.75%, respectively (Fig. 3). The HR for revision, adjusted for gender, age, ASA, and fixation method, was 0.73 (95% CI: 0.70–0.76) with CoP versus MoP and 0.83 (95% CI: 0.79–0.87) of CoC versus MoP.

Figure 3.

Cumulative 15-year revision rates with BIOLOX delta CoC and CoP versus MoP recorded in the UK’s NJR.

Cumulative 7-year revision rates with MoHXLPE, CoC, and CoHXLPE observed in the EPRD were 5.4%, 3.3%, and 4.0% , respectively (Supplementary Figure) (57).

In their most recent reports, both the Swedish and Norwegian Arthroplasty Registries showed the distribution of couplings but did not provide survival estimates for different liner and head combinations (58, 59).

An analysis of data from the Danish Hip Arthroplasty (60), excluding conventional polyethylene, suggested cumulative revision rates of 3923/85 857 (4.57%) with MoP, 505/8215 (6.15%) with CoP, and 304/4766 (6.38%) with CoC. HRs adjusted for age and gender were 1.02 (95% CI: 0.93–1.13) with CoP and 0.93 (95% CI: 0.82–1.05) with CoC.

In the Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR), age- and gender-adjusted HRs for all-cause revision with CoC versus MoHXLPE and CoHXLPE versus MoHXLPE were 1.00 (95% CI: 0.96–1.04) and 0.79 (95% CI: 0.74–0.84), respectively (Supplementary Figure) (61).

The New Zealand registry stratifies the risk of revision for different THA couplings (62). Revision rates per 100 component years were 0.47 (95% CI: 0.44–0.51) with CoC, 0.56 (95% CI: 0.53–0.59) with CoP, and 0.63 (95% CI: 0.61–0.65) with MoP. Focusing on cross-linked polyethylene liners, revision rates per 100 component years were 0.46 (95% CI: 0.43–0.49) with CoP and 0.49 (95% CI: 0.46–0.51) with MoP.

The Dutch Arthroplasty Register (LROI) specifies revision rates for acetabular components in a competing-risk and raw Kaplan–Meier model (Table 3) (63). Competing risk means that, for example, a patient who dies 5 years after index surgery with a stable and clinically inconspicuous THA obviously cannot contribute to 10-year revision estimates.

Table 3.

Acetabular revision rates stratified for cross-linked polyethylene and ceramic liners in the Dutch Arthroplasty Register (LROI) (63)

Socket bearing	n	10-year revision percentage		13-year revision percentage
		Competing risk	Kaplan–Meier	Competing risk	Kaplan–Meier
HXLPE	160 639	1.6 (1.5–1.6)	1.9 (1.8–2.0)	1.8 (1.6–1.9)	2.1 (2.0–2.3)
Ceramics	23 401	1.6 (1.4–1.8)	1.8 (1.6–2.0)	1.9 (1.6–2.2)	2.2 (1.9–2.6)

Cumulative revision rates because of infection

Infection was studied by 15 experimental and observational studies (38, 39, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76) enrolled in previous meta-analyses.(77, 78) Raw cumulative infection rates were 97/17 633 (0.55%, 95% CI: 0.45–0.67%) with CoC, 1351/158 341 (0.85%, 95% CI: 0.81–0.90%) with MoP, and 6/811 (0.74%, 95% CI: 0.27–1.60%) with CoP. Estimates from random-effects meta-analysis of proportions were 0.44% (95% CI: 0.10–0.93%), 0.59% (0.26–1.01%), and 0.21% (0.00–1.04%), respectively.

An analysis of the UK’s NJR (56), published in 2018 (79), showed cumulative incidences of revision for prosthetic joint infections of 0.410% with MoP (1/244), 0.342% with CoP (1/292), and 0.343% with CoC (1/291). With marginal and even zero event rates, random-effects meta-analysis of risk differences favored CoC pairings, mainly driven by comparisons of CoC with MoP.

The high-granularity commissioned NJR data exploration revealed that, after adjustment for gender, age group, ASA class, as well as stem and cup fixation, the observed expected ratio of infections was 0.88 with alumina–zirconia CoC and 1.04 with MoP (P < 0.001).

Ceramic fractures

Unadjusted implant fracture rates observed in the UK’s NJR are listed in Table S7 as well. After adjustment for gender, age group, ASA class, as well as stem and cup fixation, O:E ratios of head and socket fractures were 1.55 and 1.71 with BIOLOX delta CoC and 0.62 and 0.44 with MoP. With CoP and MoP, O:E ratios were 1.70 and 1.12, and 0.72 and 1.00, respectively.

Squeaking and other noise

There is currently no standardized recording or reporting scheme for noise across registries. According to data from the New Zealand Joint Registry collected between 1999 and 2015, the incidence of revision for CoC hips due to noise was 24/5587 (0.43%, 95% CI: 0.28–0.64%) (80). With an average follow-up of 6 years, raw numbers from 21 cohort studies (42, 43, 44, 46, 47, 48, 50, 52, 53, 54, 55, 73, 81, 82, 83, 84, 85, 86, 87, 88, 89) and four RCT(38, 39, 65, 68) suggest an incidence of squeaking of 360/5658 (6.36%, 95% CI: 5.74–7.03%) and 9/321 (2.80%, 95% CI: 1.29–5.26%). Estimates from random-effects meta-analysis of proportions are 4.70% (95% CI: 2.27–7.89%) and 1.64% (95% CI: 0.04–4.67%), showing substantial heterogeneity (Supplementary Figure). A Cates plot (90) using natural frequencies (Supplementary Figure) illustrates how results may be put into context and interpreted reasonably.

Association between auditory phenomena and fractures with ceramic pairings

Variance-weighted regression analysis based on 21 studies with 5079 hips, which reported both the incidence of squeaking and ceramic fractures, showed high variability in the incidence of squeaking (0.0–18.5%) but low variability in the rate of ceramic fractures (0.0– 2.3%) (Fig. 4). The regression coefficient does not suggest an association between squeaking and ceramic fractures (β = 0.01, P = 0.612).

Figure 4.

Association between the frequency of ceramic fractures and squeaking.

Function, PROM, and health-related quality of life

Many studies included in this best evidence synthesis commenced when patient-centered outcomes were either not demanded or assessed by protocol or handled in a secondary fashion. In addition, researchers employed tools established in their individual country (HHS, Oxford Hip Score, WOMAC, SF-36, EQ-5D, etc.) with distinct norm values, MCID, etc. The most common instrument across investigations was the HHS. Data from 24 studies (38, 39, 40, 42, 43, 45, 46, 47, 48, 49, 51, 52, 53, 54, 55, 65, 68, 81, 83, 88, 89, 91, 92, 93) indicated a clinically relevant average functional improvement of 42.4 (range: 30.7–54.0) points in HHS from baseline to latest follow-up, without marked differences between bearing materials (Supplementary Figure).

Discussion

General findings

In this best-evidence synthesis, we attempted to identify and compile all available experimental and observational clinical scientific information on 4th-generation alumina–zirconia ceramic pairings in elective THA to provide unbiased estimates of their distinct advantages and potential risks, published until May 2023.

The title of this work is a play on words and refers to the famous quote, ‘May the force be with you’ from the Star Wars universe.

Overall revision rates favored CoC and CoHXLPE over MoP during long-term (>10 years) follow-up. Infections occurred in 0.39%, 0.33%, and 0.43%, respectively. The weighted incidence of squeaking with ceramic pairings was three times lower under controlled experimental conditions than in an observational environment.

In health economy, there is always a trade-off between distinct advantages and disadvantages when comparing one intervention to another. In this scenario, alumina–zirconia ceramic compounds are known for being associated with squeaking and fractures but may prevent deep infections in the long run. The key question is: How many alumina–zirconia ceramic THA must be implanted to avoid one deep infection, and how many events of squeaking or even fractures shall be accepted to make alumina–zirconia ceramic hips the dominant treatment strategy over MoP, CoP, etc. Our results may serve as the robust source for consecutive risk–benefit and cost-effectiveness analyses, and data will happily be shared with other researchers upon request.

Randomized controlled trials in the context of real-world evidence

Without a doubt, there is a lack of large-scale RCTs comparing alumina–zirconia ceramics with alternative couplings in THA. Marginal (adverse) event rates with approved products and procedures hamper designing feasible superiority or non-inferiority clinical trials to evaluate step innovations along the IDEAL framework (94, 95) – tremendously high costs and difficulties in achieving completeness of follow-up over years notwithstanding.

There is a common belief (or misbelief) that data from large-scale joint arthroplasty registries may replace RCTs in this scenario. However, RCTs and registries aim at different goals and are not interchangeable. RCTs investigate whether a certain treatment is effective compared to a certain control in a controlled experimental set-up. This even applies to pragmatic RCTs (96). Registries depict the efficiency of interventions and procedures, i.e. what remains of an experimentally proven effect with a certain intervention under daily care conditions (‘real-world’ evidence (97)).

Registry-based RCTs may combine the best of both worlds (98, 99). This compelling design involves the random allocation of patients to different treatments at the point of care (i.e. the experimental part) and routine longitudinal documentation along with mandatory registry follow-up (i.e. the real-world part). Distinct advantages of this approach are minimal administrative efforts, costs, and dropouts or losses to follow-up. However, it requires an established national registry culture, specific data safety regulations, and both acceptance and awareness of patients and providers for utilizing routine data for research purposes. Registry-based RCTs are currently underrepresented in orthopedics and have mainly been planned and conducted in Scandinavia and the UK (100, 101, 102)

Limitations

While the REporting of studies Conducted using Observational Routinely collected health Data (RECORD) statement (103) was developed as guidance for reporting registry data, there are currently no uniform requirements for designing, maintaining, and analyzing joint replacement registries. Completeness of follow-up and unequivocal operationalization of both exposure and outcome variables minimize the risk of misclassification and partial verification bias, which is a prerequisite for detecting rare and even rare adverse events (the most valuable and unique benefit of population-based registries). For example, the EPRD (57) is supported by the Federal Health Insurance Fund, facilitating and guaranteeing longitudinal data integrity.

Major joint replacement registries record the item ‘ceramics’ for ball and socket composition but infrequently differentiate between material generations or manufacturers. This hampers deep modeling of the contribution of 4th-generation alumina–zirconia pairings on overall component survival, revision for infection, etc.

Certain adverse events and reactions, such as noise (specifically squeaking) as well as ceramic fractures, are currently not assessed in a consistent and quantifiable manner and are also not routinely recorded in most joint replacement registries. In addition, follow-up times for some ceramic pairings are still insufficient to draw conclusive inferences about their potential value.

As squeaking and other noises with ceramic THA were not of significant scientific and media interest until the mid 2000s (104), they were not part of the original FDA IDE trials and only partly queried during later follow-up investigations. Whereas it is undebatable that patients may experience noise phenomena with ceramic hips, little if any, clinical consequences (e.g. surgical revision) were reported in clinical and population-based studies. The high variability in the reporting of noise stands in marked contrast to the low incidence of ceramic fractures, and no association between both events could be determined by regression analysis.

Biomechanical surrogates like wear, abrasion, friction, etc., were, at best, infrequently assessed and illustrated by RCTs and cohort studies, and not addressed by registries at all. Although comprehensive in many ways, this work lacks sufficient granularity to explain construct collapse by tribological malfunction. Linking advanced imaging studies and retrieval analyses to registry data may help to better understand the reason for component failure currently subsumed under unspecific terms like loosening in the future.

Conclusion

According to the current best evidence, it is between 3 and 85 times more likely that 4th-generation alumina–zirconia oxide ceramic pairings like BIOLOX delta avoid revisions for infection than causing revisions for audible noise or ceramic fractures (105, 106, 107, 108).

The results of this work may provide the basis for a subsequent health-technology assessment report, incorporating formal health-economic analysis of tangible and intangible costs and gains. They may also lead to evidence-based reconsideration of current practice in countries where the relative frequency of alumina–zirconia ceramic couplings remains low or declined during the past years – which is at odds with scientific clues. Even in the absence of a large-scale multicenter RCT, the data presented here may be considered the state-of-the-art guidance for regulatory bodies, healthcare authorities, payers, and other stakeholders for choosing effective, efficient, and durable tribological pairings in elective THA. To promote the intriguing registry-based RCT idea, operators of established and future registries must agree upon common standards and procedures, minimal datasets, and follow-up regimens.

ICMJE Conflict of Interest Statement

DS and NK received personal research grants from CeramTec GmbH, Plochingen, Germany, to conduct this project, unfolded to and approved by their employers. The sponsor had no influence on methods, data analysis, or interpretation of results of this work. All other authors (i.e. CP, MM, LZ, and AE) disclose they had or have clinical, scientific, or commercial relationships with CeramTec in the past or ongoing collaborations, none of which had any impact on this article. LZ is an Associate Editor at EFORT Open Reviews, and was not involved with the peer review process for this article. There are no other conflicts of interest relevant or related to this submission.

Funding Statement

This work was partly funded by CeramTec GmbH, Plochingen, Germany.

Data sharing agreement

The authors would be happy to share anonymized aggregated datasets with other researchers on request following current data safety and protection rules such as EU-GDPR.