Gehrke v Barrack: Ceramics Decrease the Incidence of Peri-Prosthetic Joint Infection

This week’s Orthopaedic Crossfire® debate was part of the 19th Annual Current Concepts in Joint Replacement® (CCJR®), Spring meeting, which took place in Las Vegas. This week’s topic is “Ceramics Decrease the Incidence of Peri-Prosthetic Joint Infection” For is Thorsten Gehrke, M.D. – ENDO-Klinik, Hamburg, Germany. Opposing is Robert L. Barrack, M.D. – Washington University School of Medicine, St. Louis, Missouri. Moderating is Daniel J. Berry, M.D. – Mayo Clinic, Rochester, Minnesota.

Dr. Gehrke: We are talking about peri-prosthetic joint infections (PJI). Many, many risk factors, like operation duration, age, gender, etc. contribute to PJI. And, there is this question about implant material.

During the International Consensus Meeting on PJI in 2013, we came to the conclusion that metal-on-metal bearings may be associated with the higher risk of PJI, which is almost proven. But nothing about ceramics.

In a recent systematic review and meta-analysis about the bearings (Hexter, et al., BJJ, 2018), they could not find any statistically significant differences between the bearings.

In the Australian Joint Replacement Registry there is no difference between ceramic-on-poly and metal-on-poly regarding infection rate.

The Danish Registry shows a much lower rate of peri-prosthetic infection with ceramic-on-ceramic when compared to metal-on-poly.

Medicare data showed the same according to the works of Bozic, et al. (CORR, 2010; JOA, 2012). And a paper from Kurtz, et al., showed a statistical significance based on more than 315,000 Medicare total hip replacements that a ceramic bearing surface, whether on ceramic or poly, versus metal-on-poly has a much lower infection rate (JOA, 2016).

Jay Parvizi, M.D. conducted a study at the Rothman Institute comparing the PJI rate for metal-on-poly versus ceramic-on-poly and they found an almost statistically significant difference between the patient groups (p=0.04).

Robert, you might say, “Yes, of course. Ceramic-on-poly patients are younger and healthier.” So, the Rothman Institute study did a multivariate analysis controlling for age, Charlson comorbidity, BMI, etc., and they found out that a metal-on-poly bearing surface is an independent risk factor correlating with higher incident of PJI (Smith, et al., EAO, 2014). The use of a cobalt-chromium femoral head was an independent factor in causing PJI.

If we look at the international registry data – a small registry from Slovenia in Europe shows a much lower infection rate.

If you go to the bigger countries like the UK, the same—not significant—but still a lower infection rate in favor of ceramic-on-poly.

And if you put many registries together, 10 registries with more than 1.1 million patients, an almost statistically significant difference between them is reported. In New Zealand you find the same.

But what is reason? The reason could be that the material does not allow the formation of biofilm on its surface (Peters, Dtsch Ärztebl, 1988).

So, what is important if we develop a peri-prosthetic infection? Type of biomaterial can influence bacterial adhesion, like roughness, pH, molecular composition, ionic strength. And we also know from the literature that certain surface treatments have been described to promote osteoblast adhesion or to inhibit bacterial adhesion.

It’s proven and it’s published that adverse local tissue reaction creates an environment for bacteria growth. There is published data supporting this theory where they cultured S. epidermidis and S. aureus on different surfaces and concluded that biofilm formation is much lower on ceramic surfaces than on metal surfaces (Rimondini, ISTA, 2015).

In summary, registries suggest, not very significantly, that ceramics are associated with lower risk of infection in total hip replacement. The first in-vitro and ex-vivo results show lower bacterial biofilm adhesion on ceramic bearing surfaces.

Dr. Barrack: The thing is…we really don’t disagree because your concluding slide said that ceramics are associated with a lower incidence of infection. And they are associated with it. But, especially with registry data, associations are easy to show; causation is not so easy.

Ceramics don’t cause a lower infection rate, they’re just associated with a lower infection rate. I’ll show you the danger of not understanding the differences.

In the Midwest, on the farms every morning the rooster crows and the sun comes up, so they are associated. But most of us don’t think that the rooster controls the sun coming up.

So, to say that they decrease PJI is inaccurate because it implies causation. There is no causation, just an association.

So, what is the data? Looking at big registry sets it’s easy to find associations. It’s like quoting the Bible—you can find any association you’re looking for, but the associations are weak. When you’re up to 100,000 and your hazard ratio is only 1.2-1.5, it really is sort of meaningless.

In fact, these small differences are easy to explain by differences in the patient population.

Six months ago, at AAHKS [American Association of Hip and Knee Surgeons], using data from the New Zealand Registry—the same registry—Smith, et al. reported that patient specific differences—like comorbidities—were 2-6 times greater than the hazard ratio of infection decrease with ceramics. Every known risk factor is more likely to be present in the non-ceramic cohort.

Another major flaw in the data is coding accuracy, particularly of complications in registries. A large review article from the Journal of AAOS showed the complication coding is off by 20% (Patel, et al., 2016). Ten years ago, at the AAOS Annual Meeting, Froimson, et al., showed that coding of infections was off by about 40% at the Cleveland Clinic compared to what the surgeon’s thought was an infected case.

Until very recently, most trunnion cases were coded as infections. This could account for the apparent association with a higher infection rate. There’s no doubt that most of us believe strongly that there is less taper damage with a ceramic head. So, that’s a good reason. That’s why I use ceramic heads, just as you heard described. But not to lower the infection rate, just to lower the trunnionosis rate, which has been misdiagnosed as infection a high percentage of the time (Kurtz et al., CORR, 2013).

Many of the articles Thorsten showed you, every controlled trial where they did try to control for several confounding variables showed no difference infection, revision…only differences were in squeaking. In the large Cochrane reviews (BMJ, 2011), the results didn’t show any advantage for ceramic-ceramic compared with traditional metal- or ceramic-on-poly.

The Nordic Arthroplasty Register Association also showed no difference in revision rate except for a higher revision rate for fracture.

The other big issue is the confounding variables. Specialty procedures like ceramics tend to go to surgeons who have better patient mix and have lower infection rates. The failure to control for these confounding variables can lead to conclusions that are at odds with the literature and even the common sense.

In conclusion, I think registry data is invaluable. But it is very poor if it’s distinguishing small differences in well-performing devices. The vast majority are association and it leads to misuse.

Our great philosopher, Mark Twain, from Missouri up the river from St. Louis, said “There are three kinds of lies: lies, damned lies, and statistics.”

But my favorite quote from him is “Politicians and diapers must be changed often, and for the same reason.”

So, Thorsten, I was interested in the data you presented but to me it may be time to change that diaper.

Moderator Berry: Thorsten, the associations might be explained by both selection factors that aren’t adequately controlled for or misdiagnosis of a few taper corrosion cases for infections. So, is it possible that Robert’s argument holds water or are you going to dispute it?

Dr. Gehrke: You’re absolutely right and Robert is right. I also have my criticisms regarding the registries, they are sometimes misleading. On the other hand, what objective data do we have other than registries? We cannot really ignore the registries. I know, there are many confounding factors. We should do a multi-variate analysis and exclude all the confounding factors like BMI, co-morbidities, so on. We need more robust data. And we have to look at the surface properties and formation of biofilm. That would be the most objective.

Moderator Berry: Robert, will you concede that there is a risk for taper corrosion if you do not use a ceramic head? If you use a cobalt-chrome head?

Dr. Barrack: Absolutely. I’ve used nothing but ceramic for years.

Moderator Berry: That’s fine. Will you concede that if you get taper corrosion—admittedly it’s a 1% problem area— and if you get a bad adverse local tissue reaction, that patient might be at a little increased risk for infection.

Dr. Barrack: That would be a good topic for an in-depth study. I think that 90% of the time it would probably be a false positive.

Moderator Berry: Sometimes you do know if there are positive cultures, right?

Dr. Barrack: Right. Have you seen that?

Moderator Berry: Yeah, we have. It does happen. The question is does it happen more often than just any other revision.

Dr. Barrack: But I approach that by just doing ceramic heads. I think that Thorsten’s point brings up the bigger topic about how we interrupt registry data. When you’re talking about failure rates of 3, 4, 5, 6% higher, registries are good. When you’re looking at hazard ratios of fractions of a percent the chances that you have confounding variables are more likely,

Moderator Berry: Your point is very well taken. The ability to adjust for confounding variables is just plainly never very good. Thorsten, the data does look as though ceramics probably are a little more resistant to formation of biofilm than metal. But we can’t say in hip replacement that you’re getting rid of the metal. There is still metal in there, even if you use a ceramic head and a ceramic bearing. Does getting rid of one-third of the metal really make a difference?

Dr. Gehrke: To be honest, no idea. But I think it can influence it a little. We need some more research about it, but I think it influences a little.

Moderator Berry: Gentlemen, both thank you for a stimulating and thought provoking discussion.

Please visit www.CCJR.com to register for the 2019 CCJR Winter Meeting, – December 11 – 14 in Orlando.