Anti-CCP vs rheumatoid factor

What the research found.

• One review pooled the studies of anti-CCP, where it caught 67 percent of the cases, between 62 and 72. It cleared 95 percent of the people who didn't have it, between 94 and 97. A positive result came with a likelihood ratio of 12.46, between 9.72 and 15.98, and a negative one with 0.36, between 0.31 and 0.42.

  Nishimura and colleagues, Annals of Internal Medicine, 2007

• In that review, IgM rheumatoid factor caught 69 percent of cases, between 65 and 73, and cleared 85 percent of the people who didn't have it, between 82 and 88. So it catches a shade more than anti-CCP does. It's also wrong far more often when it comes back positive.

  Nishimura and colleagues, Annals of Internal Medicine, 2007

• One review covered 151 studies of anti-CCP, and the spread across them is the point. The share of cases caught ran from 12 to 93 percent, and the share of healthy people cleared ran from 63 to 100 percent. So a pooled figure hides how much the tests differ from each other.

  Whiting and colleagues, Annals of Internal Medicine, 2010

• In early rheumatoid arthritis, study design changed the answer. Studies that followed a group forward gave an anti-CCP catch rate of 54 percent, between 48 and 60, while two other designs gave 66 percent, between 59 and 72. So the test does worst when it's needed most.

  Whiting and colleagues, Annals of Internal Medicine, 2010

• Two databases followed rheumatoid arthritis patients from 1980 to 2004, and the first recorded measurement is the telling one. More than a third had one normal or negative result. That meant a normal sedimentation rate, a normal C-reactive protein, or a negative rheumatoid factor, in people who unmistakably had the disease.

  Sokka and colleagues, The Journal of Rheumatology, 2009

The short answer.

Anti-CCP is the more useful test, and the reason is how often a positive result is right rather than accuracy in general. Both tests find about as many people who have rheumatoid arthritis, at 67 percent against 69 percent, which is a difference of no consequence. Where they part company is the false alarm rate.

Anti-CCP cleared 95 percent of the people who didn't have the disease, and rheumatoid factor cleared 85 percent. In practice that means a lot, because a positive anti-CCP multiplies the odds you have rheumatoid arthritis by about twelve while a positive rheumatoid factor moves them much less. Rheumatoid factor turns up in Sjogren disease, in hepatitis C and other infections, and in healthy older people.

Neither test rules the disease out. About a third of people with rheumatoid arthritis are negative on either, which is what seronegative disease means, and it means the disease with negative blood tests. It's common enough that a diagnosis rests on the whole clinical picture rather than on a result.

Side by side.

What each test is looking for

Anti-CCP looks for antibodies against citrullinated peptides, which are proteins that have gone through a specific chemical change. Those antibodies are closely tied to rheumatoid arthritis, and they turn up in fairly few other conditions. That property is what does the work in this comparison.

Rheumatoid factor is an antibody aimed at part of your own antibodies, and it was described decades earlier, which is how the disease got part of its name. It's a good deal less fussy about which condition it appears in. That's where its weakness comes from, and it's the whole of the difference.

Both get ordered together in most places, so a reader looking at a results page sees them side by side. Nobody tells her that one counts for more than the other. It does, and the difference is worth understanding, so the rest of this page is about where it comes from.

Catching cases, clearing healthy people, and which one counts here

One number is the share of people with the disease that a test finds, which doctors call sensitivity, and on that measure these two are level. Anti-CCP caught 67 percent, between 62 and 72, and rheumatoid factor caught 69 percent, between 65 and 73. So if that number decided it, there would be no comparison to make.

The other number is the share of people without the disease that a test leaves alone, which doctors call specificity. Anti-CCP cleared 95 percent, between 94 and 97. Rheumatoid factor cleared 85 percent, between 82 and 88. Ten points sounds modest, and it isn't, because it changes how much a positive result should move your thinking about the whole question.

The likelihood ratio puts that into one figure. A positive anti-CCP comes with a likelihood ratio of 12.46, between 9.72 and 15.98, which means it multiplies the odds of rheumatoid arthritis by about twelve, and a negative one comes with 0.36, between 0.31 and 0.42. Rheumatoid factor clears fewer healthy people, so it moves the odds a good deal less, and that's the practical difference between the two tests.

Where both of them fail

Neither test rules the disease out, and the arithmetic runs the same for both. A catch rate around two thirds means roughly a third of people with rheumatoid arthritis test negative, which is what seronegative rheumatoid arthritis means, and it means the disease with negative blood tests. It's common rather than odd, and it's worth knowing the word.

The problem is worse early on, which is just when a test would be most useful. One review looked at early disease, where studies that followed a group forward put the anti-CCP catch rate at 54 percent, between 48 and 60, while two other designs gave 66 percent, between 59 and 72. So a test that finds half of early cases can't be relied on to rule anything out.

The inflammation markers make the point a second time. Two databases followed consecutive patients from 1980 to 2004, and at the first recorded measurement more than a third had one normal or negative result, meaning a normal sedimentation rate, a normal C-reactive protein, or a negative rheumatoid factor. So normal blood work in rheumatology isn't the reassurance it looks like.

Why the numbers on your report may not match these

The pooled figures on this page are averages, and they come from a large and varied literature. One review covered 151 anti-CCP studies, where the catch rate ran from 12 to 93 percent and the clearing rate ran from 63 to 100 percent. That spread reflects real differences, because the tests differ, the people studied differ, and the yardstick each study compared against differs.

So a single result sits inside a wider band of doubt than a printed number suggests. A strongly positive result and a borderline one aren't the same finding. A borderline result is worth discussing with the person who ordered it rather than acting on by itself.

It also means a comparison like this one describes the tests in general rather than the test you had. Your rheumatologist knows which test the local lab runs, and this page can't know that. So take a puzzling result to them rather than to a website.

When each one fits.

• Anti-CCP is the test worth having when the question is whether joint symptoms are rheumatoid arthritis, because a positive result counts for the most.
• Rheumatoid factor still gets ordered alongside it, since a positive result in both is stronger than either alone, and it has a role in other conditions.
• A negative anti-CCP shouldn't close the question, which holds most in somebody with swollen, stiff joints on both sides, because seronegative disease is common.
• In very early disease, neither test performs as well as its pooled figure suggests, which is an argument for a rheumatology opinion rather than a repeat test.
• Neither test gets followed over time toward a target, so repeating one to see whether it improves misunderstands what it does.

Questions patients ask.