JOA - 2026-09-07 - Journal Article
Reducing the Surveillance Burden in Metal-on-Metal Hip Arthroplasties: A Risk-Adapted Three-Tier Model Based on a Large Patient Cohort.
Selim A, Saeed A, George A, Clarke M, Bonde S, Naseem S, Crooks S, Dickenson E, Nandra RS, Thomas G
Topics
Key Takeaway
A three-tier risk-adapted MoM surveillance model achieved sensitivity of 0.84 and AUROC of 0.67 while reducing annual follow-up burden from 69% to 53% of patients and saving £46,000–£59,000 per 1,000 patients annually compared to the MHRA protocol.
Summary Depth
Choose how much analysis to show on this article page.
Summary
This study asked whether a risk-stratified three-tier model could safely reduce MoM hip surveillance burden compared to the 2017 MHRA mandatory protocol. Logistic regression on 845 MoM resurfacings and large-head THAs identified Tier-1 predictors (female sex aOR 2.0, bilateral aOR 1.7, GFR <60 aOR 1.6) and Tier-2 predictors (abnormal X-ray aOR 10.6, pain aOR 3.7, elevated metal ions aOR 2.5). The hierarchical model outperformed MHRA on specificity (0.51 vs 0.34) and AUROC (0.67 vs 0.61) while reducing annual follow-up from 69% to 53% of patients and cutting workload by 75%.
Key Limitation
The model was derived and validated in a single-center cohort without external validation, limiting generalizability across implant registries with different MoM device distributions and healthcare systems.
Original Abstract
BACKGROUND
Since 2017, the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom (UK) has mandated comprehensive surveillance for patients with metal-on-metal (MoM) hips. We aimed to develop a risk-adapted follow-up pathway, with the goal of safely reducing unnecessary visits and investigations.
METHODS
We analyzed 845 MoM hip resurfacings and large-head MoM Total Hip Arthroplasty (THA). The primary outcome was abnormal magnetic resonance imaging (MRI) or revision. Logistic regressions identified the predictors, which were translated into a clinical risk score for Tier-1 and Tier-2.
RESULTS
In Tier-1, the strongest predictors (by adjusted odds ratio, aOR) were women (2.0), bilateral (1.7), glomerular filtration rate less than 60 (1.6), age less than 50 years (1.4), head size less than 48 mm (1.2), and high-risk implant (1.2). In Tier-2, the predictors were abnormal X-ray (10.6), pain (3.7), abnormal metal-ion level (2.5), and other symptoms (1.6). The hierarchical model, using thresholds of Tier-1 ≥ 3 and Tier-2 ≥ 2, achieved a sensitivity (0.84), a specificity (0.51), and an area under the receiver operating characteristic curve (AUROC) of 0.67. By comparison, the MHRA approach demonstrated a sensitivity (0.88), specificity (0.34), and an AUROC of 0.61. The hierarchical model reduced annual follow-up from 69 to 53%. High-risk reviews were extended to three-yearly, and low-risk surveillance was discontinued. This resulted in savings of £46,000 to £59,000 [$61,400 to $78,800] per 1,000 patients annually, a 75% reduction in workload compared with the MHRA approach.
CONCLUSIONS
We propose a risk-adapted, three-tier pathway: (1) baseline stratification into high- and low-risk groups using the clinical risk score; (2) surveillance of high-risk patients every three years, including symptom review, radiographs, and optional metal-ion testing; and (3) MRI reserved for patients who have Tier-2 abnormalities above the risk score. Low-risk patients have limited benefits from follow-up beyond 10 years unless symptomatic.