Cementless or cemented in a robotic knee: the evidence cannot say

Design

Systematic review and meta-analysis of comparative observational studies, ROBINS-I and GRADE

Population

Ten studies of cementless versus cemented primary robotic-assisted total knee arthroplasty

Primary outcome

Mid-term all-cause revision

Effect

RR 0.63 (95% CI 0.09 to 4.53); manipulation under anaesthesia RR 1.03 (0.09 to 11.74); all outcomes very low certainty

Ten comparative studies were pooled to ask whether the better bone preparation a robot provides changes which fixation to choose in primary total knee arthroplasty. The answer, honestly reported, is that nobody knows. Mid-term all-cause revision gave a risk ratio of 0.63 but with a confidence interval running from 0.09 to 4.53. Manipulation under anaesthesia was 1.03, interval 0.09 to 11.74.

Operative time looked shorter with cementless fixation by about 13 minutes, but with an interval crossing zero and heterogeneity at 94.4 per cent — meaning the studies disagree with each other more than they agree. Length of stay showed nothing either way. Certainty was graded very low for every pooled outcome, and the analysis was of observational data with ROBINS-I assessment, not trials.

The practical reading is that robotic assistance is not by itself an argument for changing fixation. Whatever decided your cementless threshold before — bone quality, age, deformity, the implant system you trust — still decides it.

• Do not adopt cementless fixation on the grounds that the robot makes the cuts more reproducible; that link is untested.
• Keep patient-level selection criteria for cementless fixation: bone stock, age, inflammatory disease, obesity.
• Audit your own revision and manipulation rates; registry-level data will answer this before pooled series do.
• Cost matters in Indian practice: cementless implants and robotic time both add to the bill without demonstrated mid-term gain.
• Treat the 13-minute operative-time claim as setting-dependent until a trial reports it.

Why it matters

It stops a technology argument being used to settle a fixation decision it has no evidence on.

Don't overread it

Very low certainty observational evidence: this shows absence of evidence, not evidence that the two are equivalent.

The statistics, in plain English

A risk ratio of 0.63 sounds like a large benefit until you read the interval: 0.09 to 4.53 spans a tenfold reduction and a fourfold increase, which means the data are compatible with almost any truth. That width comes from three studies and very few revision events. High heterogeneity on operative time (I squared 94.4 per cent) says the studies are measuring different things — different systems, teams and learning curves — so the pooled mean is not a number to plan a list around.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.