When it comes to choosing among competing products, tradeoffs are all
but inevitable. If you want quality, you’ll have to sacrifice price. If
you go for the cheapest product, quality will probably suffer. But what
if you could have the best of both worlds, and get fast service to boot?
That kind of trifecta is available in hepatitis C testing, except that
it’s not a product—it’s an algorithm, a way of triaging specimens
that test positive on an HCV antibody screening test.
But though the algorithm saves roughly $100 in testing cost for each
positive HCV patient, few clinicians are aware of it.
For the past year and a half, the Centers for Disease Control and Prevention
has promoted a set of guidelines to make HCV testing much more cost-effective
by shortening the confirmatory test process. The key to the guidelines
is moving beyond reporting the traditional HCV antibody screening test’s
binary result (positive or negative) to reporting that distinguishes between
low-positive HCV antibody results and high-positives.
As many studies have shown, the majority of specimens with low signal-to-cutoff
(S/CO) ratios are found to be negative with confirmatory testing. Using
that correlation, laboratories can more efficiently choose a confirmatory
test, stop the workup earlier, and not compromise results reporting at
The capsule version of what CDC recommends is this: Laboratories should
do a RIBA (recombinant immunoblot assay) test for low-positives, and an
RNA test for high-positives, says Edward Ashwood, MD, senior vice president
and director of laboratories at ARUP Laboratories in Salt Lake City, and
professor of pathology, University of Utah School of Medicine.
"For patient care, the challenge in HCV testing is to find a test-ordering
algorithm that will identify those currently infected with HCV in the
shortest possible term," he says. "For cost-effectiveness, the goal is
to develop a pattern that will eliminate the largest number of candidates
from progressing to each successive round of testing." The CDC-recommended
algorithm accomplishes both goals, he says.
"There are five times as many high-positives as low-positive HCV results.
Fifty-two percent of the low-positives will be RIBA-negative, and you
can stop the workup. Ninety-three percent of the high-positives will be
HCV RNA-positive, and you can stop the workup."
The CDC is just now analyzing results of a recent survey it conducted
to measure the impact of the 2003 guidelines. But reports from the field
suggest the guidelines are not widely employed, and a lot of consciousness-raising
may be needed before they are.
"I’m not sure that many people know there are guidelines," says Nancy
Cornish, MD, director of microbiology for The Pathology Center and Children’s
Hospital, Omaha. In fact, data she has collected over the last two years
show there is an even more basic information gap about confirmatory testing.
While cautioning that hers is a community hospital and other institutions
may have different findings, Dr. Cornish says, "It’s only been a few months
since we actually started our flow charts, but from the positives we’ve
had, there are enough that I’m realizing physicians aren’t clear about
what to order next or that they even have to do additional testing."
They had 2,112 total HCV tests in 2003 and 3,352 in 2004. "A total of
258 tests were positive on the serologic antibody screening test, and
of those, we had 139 patients who did not appear to have any kind of follow-up
of the screening test. There was no confirmatory test I could tell from
my data," she says.
"Then there were a number that had two confirmatory tests, like a RIBA
serologic test and a PCR, and there were a number with incorrect confirmatory
tests performed. For example, the clinician ordered a PCR confirmatory
and if it was negative, did nothing further."
That’s wrong, because if the screening test is positive and a PCR is
negative, they should reflex back to a RIBA to prove the patient actually
had antibodies present due to hepatitis C because sometimes they don’t,
she says. "But a subset of the population, maybe 10 to 15 percent, are
truly infected with HCV and will have cured the disease on their own.
They are not infectious and they don’t have any virus circulating."
Dr. Cornish’s local study was sparked by a belief that clinicians just
weren’t following up positive test results appropriately. "We were suspicious,
but we didn’t have any data. We were putting a comment on all screening
tests that confirmatory tests needed to be done, there was additional
serum available in the department if needed, and to call the laboratory
if they had questions. But we were not getting many calls."
At that time, The Pathology Center was running a test that did not report
if there was a high signal-to-cutoff ratio, or a weakly positive or strongly
positive result. "It would just answer yes or no. We’d run the test, we’d
say when it was positive, put the comment on the report, and we didn’t
do anything further," Dr. Cornish says.
Then a physician in a nearby hospital had a needle stick injury. "Per
protocol, he was tested for HCV and came up positive. No confirmatory
test was done, and three months later he was tested again and the screening
test came up positive for HCV, and a third time as well. Then the specimen
was sent for PCR and the PCR was negative, so they sent off another screening
test to a different laboratory and it also came back negative."
"We know screening tests don’t always agree with each other. Some will
come up positive and some negative, especially if the test is a low-positive
result. But this is an example of what shouldn’t have happened," she says.
After that incident, her hospital planned to switch to using the signal-to-cutoff
ratio to decide whether to do a RIBA or not. "Then I did that data collection
of all the patients and realized a significant number were not getting
followed up." Most of them were positive outpatients, and not specifically
liver patients, she adds.
Why are the screening tests ordered? "Some are ordered for sure because
the patient has elevated liver functions," Dr. Cornish says, noting that
most of the ordering physicians are primary care providers. "Others may
be ordered because the patient himself is concerned, or has a history
of IV drug abuse or of being in contact with somebody with HIV."
But the most significant problem is failure to order confirmatory testing,
Dr. Cornish maintains. In the past, "We left it up to the clinician. We
put a comment on the report that they needed it but left it up to them
to order it. And I quite frankly assumed they knew what to do and if they
didn’t know, they would give us a call."
When Dr. Cornish talked to the laboratory manager of the hospital where
the doctor had the needlestick, however, "She said she didn’t think they
even read the comment. She was adamant we needed to do reflex confirmatory
testing—to the point where she said she would not use us as a reference
laboratory if we didn’t."
Nevertheless, Dr. Cornish’s laboratory is still not technically performing
reflex testing at this point. "If we have a low-positive, it’s very easy
to send the serum for RIBA confirmation, and we already have that serum
here so it’s no problem," Dr. Cornish says. "But if the patient is a high-positive,
I have to call the clinician and talk to them and find out what’s going
on, then recommend an additional specimen be drawn for PCR reflex to genotype
depending on the circumstances."
"So far I have found a number of cases where reflexing automatically
to PCR was not necessary. For example, we had a patient with a high-positive
who told the doctor he had already been treated and cured, so the physician
ordered another screening test, which was probably not a good thing to
do because it will come back positive for life. Maybe he was trying to
document the case, but it costs money."
"So what should I do with this result? I don’t think I should haul off
and do PCR, which is expensive—more than a hundred dollars, depending
on the hospital. So what I wound up doing was putting a big comment on
the report saying I had a consultation with the patient’s primary physician
and his hospital physician, and I didn’t actually do a reflex test but
left it up to them if necessary."
Believing that it’s important for laboratories to take the educational
initiative on appropriate HCV confirmatory testing, Dr. Cornish, with
the support of her pathology director, Thomas L. Williams, MD, has launched
a campaign at her hospital to get the word out, using online news bulletins,
workshops, and conference presentations. (Related article: Reflex
hepatitis C testing protocol.)
Incorrect HCV testing is performed not only at community hospitals, she
emphasizes. "I was thinking university hospitals would be more up on the
guidelines, but I’ve been told it is a problem everywhere."
"Academic institutions can use cost-inefficient algorithms as well as
anyone," ARUP’s Dr. Ashwood agrees. A few of the better-run private hospitals
may be more attuned to efficiently conducting HCV testing because they
are looking more closely at costs, he suggests.
"But a lot of people don’t know about these recommendations," Dr. Ashwood
continues. "I delved into my database to see how often these tests were
ordered appropriately, and people are clearly not following a cost-effective
algorithm. I’d say 60 percent of RIBA tests are not needed."
Each of those extra tests costs about eight times as much as the antibody
screen, he adds, so costs would drop substantially, along with test volume,
if a more cost-effective testing workup were used.
D. Robert Dufour, MD, is the author of a 2003 study that supported the
CDC guidelines, and he was a member of the advisory panel that helped
develop the guidelines. A consultant in pathology at the Veterans Affairs
Medical Center, Washington, DC, and professor of pathology at George Washington
University Medical Center, Dr. Dufour helped start the VA on the new testing
algorithm in 2000. The adoption rate has been slow. Only about 15 percent
of VA hospitals are reflexively doing confirmatory testing, he estimates.
"I talk about this a lot. I’m invited to speak on HCV quite a bit and
I try to publicize this every place I go, but we’ve not seen a lot of
adoption even within the VA. The VA came out with an advisory to laboratories
suggesting that ideally confirmatory testing should be done this way,
but it gave laboratories the choice. Not doing it was acceptable, and
the laboratory could just allow the physician to decide."
In the VA setting, the number of false-positive HCV tests tends to be
rather high, Dr. Dufour notes. "We found about 20 percent of our positive
results were weakly positive, and almost 90 percent of those were false-positive,
so about 18 percent of total positive results were false-positives." While
the same is probably true for HIV antibody tests, 18 percent is higher
than the false-positive rate for most laboratory tests.
Whether confirmatory tests are ordered depends on the category of clinician,
he adds. "Most people who are liver specialists would know about confirmatory
tests, but internists or general practitioners probably don’t," says Dr.
Dufour. "And they would be doing the majority of HCV testing."
In his work as attending physician in a liver disease clinic, "Before
we started reflex testing, a lot of people were referred to us who had
never had confirmatory testing. We’d see no risk factors for HCV, normal
liver enzymes, and low values on the HCV antibody test. We’d do confirmatory
testing and find they were negative and have to reassure them."
It was the effect of a false-positive on an individual patient that was
the impetus for doing confirmatory testing. "That person may have already
been referred to a liver specialist, been labeled as having HCV, and maybe
had to make changes in lifestyle or may have difficulty getting life insurance,"
Dr. Dufour says. "I’ve personally written letters to insurance companies
saying we’ve investigated and this is a false-positive, but still the
company refuses to insure them."
If a patient has a screening test that’s positive, to fail to get a confirmatory
test is a tremendous disservice, says Tom Safranek, MD, a public health
epidemiologist with the Nebraska Health and Human Services department.
Nevertheless, it’s an omission that is still common. "I suspect those
CDC recommendations are falling on deaf ears right now," Dr. Safranek
Galvanizing the physician community on HIV diagnosis, by comparison,
is simple, because if it is not correctly diagnosed and treated, there
is a predictable bad outcome for a large percentage of patients. "With
HCV, one challenge is that a high percentage of people who are infected
can live a normal life span and die from some cause unrelated to HCV,"
Dr. Safranek says. "It can contribute to a kind of complacency that’s
not there for HIV infection."
In Dr. Cornish’s view, many problems could be avoided if hospitals maintained
a portable electronic medical record. "I’ve had a number of patients with
high ratios that, if I’d automatically done reflexive PCR and genotype
testing, it would have been a waste of time. For example, one patient
was already diagnosed with HCV and had had both tests done, but the physician
was not aware of it; it just happened to be in our computer."
"There was a needlestick injury associated with this patient so they
went ahead and did another screening test. Because our hospital serves
a number of other hospitals in the area, I had it in our laboratory computer
but they didn’t have that information available on the floor." Another
patient had a negative PCR test six months earlier that the physician
was unaware of and for which a RIBA was needed to complete the algorithm.
"He turned out to have a false-positive screening test," Dr. Cornish says.
If they had a portable record, this repeat testing could be avoided, she
adds. "It’s happened to me over and over: We waste a lot of time and money
because we don’t have all that information in a retrievable form."
Another difficulty may crop up when testing is performed after treatment
has started. "We had a patient in our initial study where the genotype
was not done up front. Knowing the genotype is essential because it determines
length and type of medication for treatment. They started treating the
patient, then tried to go back and do the genotype, but couldn’t because
the viral load had already dropped. That’s a problem I hope we fix with
In Nebraska, Dr. Safranek says, the practice habits of providers haven’t
been optimized so that they’re doing as good a job with hepatitis C as
with HIV. "Frankly, one of the challenges of working in public health
is finding the resources to look at this stuff," he says. "Are our laboratories
even measuring the strength of positivity or whether it is a high or low
signal-to-cutoff? That’s the first thing we ought to be assessing, and
we haven’t collected that information."
Nebraska is not alone. To be sure, most diagnostics manufacturers have
not yet labeled their products to correspond with the CDC guidelines,
and many large reference laboratories do not report their results this
way either. (If the laboratory does not report S/CO values, the CDC recommends
performing reflex testing of all positive anti-HCV specimens with RIBA
or nucleic acid testing.)
Distinguishing between low- and high-positives is also not an exact science.
On the enzyme immunoassay, the CDC defines a low-positive range as 1.0
to 3.8 S/CO. On the chemiluminescent immunometric assay, which ARUP Laboratories
uses, low-positive results are between 1.0 and 8.0 S/CO.
But "there’s some concern about whether the calibration of the HCV antibody
assays is stable enough to reproducibly split up results into low-positive
and high-positive," Dr. Ashwood says. Though the manufacturers strive
to maintain 1.0 S/CO without drift, when reagent lots change, "there are
no guarantees of stability at 3.8 EIA S/CO and 8.0 CIA S/CO."
While ARUP’s test results agree with those the CDC reports, Dr. Ashwood
disagrees with physicians who base their diagnosis of HCV on the results
of the EIA alone. The CDC reports that for high-positive results this
strategy is adequate because it is accurate 95 percent to 98 percent of
the time. ARUP’s data show that a diagnosis of current infection would
be correct only 93 percent of the time (that is, only 93 percent of the
chemiluminescent high-positive specimens are HCV RNA-positive).
Another unfortunate side effect of false-positives is an epidemiological
issue, Dr. Ashwood says. "Frankly, we don’t have a good feel for how much
HCV is really out there, because we’re getting a lot of screening tests
reported as positive but we’re not sure how many are true positives."
The false-positive rate of HCV antibody testing can be hard to pin down.
False-positives are a worse problem in comparatively healthy populations,
so volunteer blood donors are more likely to have a false-positive than
a true positive, while the new prisoners that Nebraska screens for HCV
tend to have the reverse pattern. "If it’s an inner-city area where there’s
a high prevalence, you have fewer false-positives. Where I’m working,
I have more because I’m dealing with a low-prevalence population," Dr.
ARUP Laboratories is about to publish a new HCV testing algorithm in
line with the CDC recommendations. But even ARUP’s old algorithm was a
major improvement on early ones, because Dr. Ashwood has long sought ways
to make confirmatory testing more efficient, and in the process he has
succeeded in slashing the costs of HCV supplemental testing.
"Most of the positive screens in the early days would have positive RIBAs—then
all those people would need an HCV RNA test performed to see if they were
currently infected." That was the algorithm CDC endorsed in the 1990s,
But even before Dr. Dufour’s study on the relevance of low- and high-positives,
Dr. Ashwood reasoned that if you went straight to the RNA test you could
skip 82 percent of the RIBAs and still have two tests indicating an HCV
Using typical reference laboratory prices, the cost savings of ARUP’s
old diagnostic algorithm are remarkable. Dr. Ashwood compared testing
costs for 20,000 people with a 10 percent prevalence, and found (assuming
2,200 HCV antibody-positive test results) that a RIBA-first-then-PCR-if-indicated
strategy cost $481,000 and took at least six days to confirm a current
On the other hand, performing HCV RNA by PCR first, then anti-HCV by
RIBA if indicated, brings the cost down to $317,000. This happens because,
while there will be 2,200 PCR tests, only 400 RIBA tests will be needed.
Total savings will be $164,000 or 34 percent—and the time to confirmation
of current infection drops to no more than five days.
The low/high strategy that will be part of ARUP’s new algorithm, Dr.
Ashwood predicts, will bring the testing cost down by perhaps another
three or four percent.
Though it supports S/CO reporting, ARUP Laboratories won’t perform reflex
testing in certain situations as the CDC guidelines ask, Dr. Ashwood says,
because of the compliance implications.
"CDC has recommended you automatically do a RIBA on a low-positive. The
physician shouldn’t even have to ask." But that recommendation makes him
"I think that strategy does not pass compliance review because we would
be performing tests that were not ordered. If CMS [Centers for Medicare
and Medicaid Services] would make a recommendation, I’d do it. But I’m
not going to let CDC dictate follow-up testing." He gives clients a choice,
but very few now order the HCV screen with reflex testing.
Laboratories must take on the mission of raising awareness about appropriate
HCV testing, Dr. Cornish insists. "Physicians are extraordinarily busy.
They’re asked to see a large number of patients every day, and it’s very
hard to keep up with all the reading you need to do to keep abreast of
all the changes that are happening," she says. "As a pathologist, in fact,
I know I’m way behind. So I don’t know how anybody in primary care is
able to do it."
With testing so complex and the time crunch for everyone so pressing,
Dr. Cornish says labs and clinicians have to start working as a team.
"And in the laboratory we have to take responsibility on behalf of the
patient, and assume the need to educate our physicians." And hepatitis
C testing, she says, is a prime example of where that’s needed.
Anne Paxton is a writer in Seattle.
- Dufour D, Talastas M, Fernandez M, Harris B, Strader D, Seeff L. Low-positive
anti-hepatitis C virus enzyme immunoassay results: an important predictor
of low likelihood of hepatitis C infection.Clin
- Dufour D. Lot-to-lot variation in anti-hepatitis C signal-to-cutoff
Chem. 2004; 50:958-960.