C. difficile back with a vengeance

May 2006
Cover Story

William Check, PhD

Unlike old soldiers, old infectious diseases don’t fade away. They bide their time until they can come back in a more dangerous guise. Case in point: Clostridium difficile-associated diarrhea, or CDAD, which has persisted in hospitals at a low level over the past quarter century, only to emerge as a hypervirulent epidemic strain in many hospitals in the U.S., Canada, and Europe during the past four years.

In fact, C. difficile puts one in mind of the sea god Nereus, from whom the Greek hero Hercules had to wrest a secret. Nereus struggled to get free by shape-shifting—assuming the appearance of various animals and forces of nature—but Hercules hung on and prevailed. Clostridium difficile, too, has shifted its DNA over the decades, adapting to whichever broad-spectrum antibiotic dominates medical practice at the time. Like Hercules, infection control practitioners are hanging on. It is not clear at this point whether they will vanquish their cunning foe or whether the best they can hope for is simply to keep it in hand.

“The appearance of what appears to be a hypervirulent strain [of C. difficile] re-awakens interest in this disease organism,” says Lance R. Peterson, MD, director of microbiology and infectious diseases research in the Department of Pathology and Laboratory Medicine, Evanston (Ill.) Northwestern Healthcare. “People have kind of taken it for granted,” notes Dr. Peterson, who is also professor of pathology and medicine at Northwestern University’s Feinberg School of Medicine, Chicago. “Unfortunately, we are now seeing outbreaks of this strain with a fairly high mortality rate.”

A number of factors have led to the emergence of this new strain. Vivian G. Loo, MD, associate professor of medicine at McGill University and chief of the Department of Microbiology at McGill University Health Centre, says one factor is an older, more critically ill hospital population. Another: “Older hospitals have infrastructure problems that lead to breaks in hospital control practices.” Dr. Loo sees a parallel between the new C. difficile strain and the outbreak of SARS in Toronto hospitals a few years ago. “In one case we had a new etiologic agent and in the other an old pathogen that changed its behavior. Hospitals don’t have the capacity to respond effectively to either,” she says.

A third factor is the genetic profile of the new C. difficile strain. “To my mind, the association between this strain and more severe disease has been proven,” says L. Clifford McDonald, MD, medical epidemiologist in the Epidemiology and Laboratory Branch, Division of Healthcare Quality Promotion, Centers for Disease Control and Prevention. He cites data from Dr. Loo and her colleagues and an analysis of a large body of data from across Canada reported in March at the meeting of the Society of Healthcare Epidemiologists of America.

That work was presented on behalf of a Canadian working group by Mark A. Miller, MD, MSc, FRCPC, chief of the Department of Microbiology, head of the Division of Infectious Diseases, and chair of the Infection Prevention and Control Unit at McGill University and SMBD-Jewish General Hospital. The toxin gene profile of the new strain was “highly and significantly correlated with increased CDAD-related morbidity and mortality,” the group concluded. Says Dr. Miller, “We hadn’t fully appreciated the relation between strain differences and clinical disease before this outbreak.” Correlations had been done on a smaller scale but never on a national basis.

Finally, a critical factor is that the new strain has acquired resistance to fluoroquinolones, which have most likely selected for its emergence. “Fluoroquinolones have been very successful, maybe too successful for our own good,” says Dr. McDonald. “What I fear is that some hospitals are starting to look like animal feedlots, where a single antibiotic is given to the entire population. In some hospitals, fluoroquinolones as a class account for 40 percent or so of all antibiotics used, and perhaps as high as 60 to 70 percent in certain patient groups.”

Antibiotic restriction may be necessary to control the ongoing CDAD epidemic, suggests Dale N. Gerding, MD, associate chief of staff for research and development at Hines Veterans Affairs Hospital and professor of medicine at Loyola University School of Medicine, Maywood, Ill. Compelling data exist that rates of CDAD declined during restriction of clindamycin and cephalosporins, the antibiotics associated with CDAD in previous decades. “For the new epidemic strain we don’t yet have good data on antibiotic control strategy,” Dr. Gerding says. Rates of CDAD due to the new C. difficile strain in Canada have been halved by emphasizing traditional infection control practices. “So far they have not attacked the antibiotic use side of things,” Dr. Gerding says. “I hope some hospitals in the U.S. try that, although they would have to revise infectious disease treatment pathways to do it.”

John Bartlett, MD, contributed greatly to the recognition of C. difficile as the cause of antibiotic-associated colitis in the late 1970s. Dr. Bartlett, now chief of infectious diseases at Johns Hopkins Hospitals and professor of medicine at Johns Hopkins University, Baltimore, notes that some hospitals experienced fairly severe outbreaks of CDAD in the original epidemic. At the General Hospital in Birmingham, England, he recalls, “There were surgical wards where 10 percent of the patients had C. difficile colitis and there was a lot of bad disease, including death. So I’m not sure that what we are experiencing now is that much different.” Quoting the words of the immortal Yogi Berra, Dr. Bartlett adds, “It’s déjà vu all over again.”

Public awareness first arose of an epidemic of CDAD in the province of Quebec, particularly in Montreal and at University of Sherbrooke Hospital, through several articles published in the July 6, 2004 issue of the Canadian Medical Association Journal. According to a detailed analysis published in CMAJ the next month, by late 2002 the number of cases of CDAD had jumped four- to fivefold (increasing eightfold among those over age 65), and there was an increased proportion of cases requiring treatment in an ICU or colectomy (Pepin J, et al. CMAJ. 2004;171:466–472). Mortality tripled.

Meanwhile, surgeons at the University of Pittsburgh Medical Center had observed an increase in the incidence of fulminant CDAD and in the number of colectomies, particularly among transplant patients (Dallal RM, et al. Ann Surg. 2002;235:363–372). Infection control practitioners analyzed the outbreak; the analysis showed that fluoroquinolone exposure was an etiologic factor (Muto CA, et al. Infect Control Hosp Epidemiol. 2005;26:273–280).

In 2004–2005 Dr. McDonald and colleagues did two online surveys about six months apart among infectious disease clinicians. Among 530 physicians who responded to the two surveys, 30 percent to 40 percent reported a perceived increase in the number of cases of CDAD or its severity, with a similar response in the two surveys. “What we took away was that some doctors were seeing it but had not tuned in to it,” Dr. McDonald says. “It was not just brought on by the news from Canada.”

Dr. Gerding has been working on C. difficile for the past 25 years. As hospitals around the U.S. experienced outbreaks and severe cases of CDAD, they sent him isolates for typing. “We discovered that hospitals in diverse parts of the country were having outbreaks with the same group of organisms,” Dr. Gerding says. When he talked to workers at the CDC, he found they were detecting organisms of the same group, called BI. In addition, researchers at the CDC had noted changes in the bacterium’s toxin genes, particularly a deletion in a putative repressor gene for toxins A and B (tcdC) and the presence of a distinct toxin called binary toxin. They also detected fluoroquinolone resistance. In his collection of about 6,000 C. difficile isolates, Dr. Gerding found only 14 examples with the tcdC deletion and the binary toxin from the 1980s and early 1990s (McDonald LC, et al. N Engl J Med. 2005;353:2433–2441). So the strain was not new, but it had previously been nonepidemic. “The only truly new feature present in the epidemic strain was acquisition of resistance to gatifloxacin and moxifloxacin,” Dr. Gerding says.

“It is possible that what was potentially a highly virulent strain was not able to get traction in hospitals until it acquired broad fluoroquinolone resistance,” Dr. Gerding suggests. Why did the virulent strain not emerge in the late 1980s and early 1990s when the earliest fluoroquinolones—ciprofloxacin and levofloxacin—were introduced? “C. difficile has an interesting susceptibility pattern,” Dr. Gerding says. “Almost all strains are resistant to ciprofloxacin and levofloxacin and almost none are resistant to gatifloxacin and moxifloxacin.” He proposes that when gatifloxacin and moxifloxacin resistance were acquired, it raised resistance to all fluoroquinolones, or FQs.

Dr. Loo and colleagues verified the presence of the new strain in Canada and its FQ resistance (Loo VG, et al. N Engl J Med. 2005;353:2442–2449). “There are many ways that strain could have appeared” in several places in the same period, Dr. Loo says. One possibility: The toxigenic strain was present in many places and mutated to FQ resistance independently many times under selective pressure as newer FQs came into widespread use. “There is some evidence for that,” Dr. Loo says. “Different historic isolates of the same type strain differ in their resistance pattern.” Further genetic analysis will answer this question.

“Use of the fluoroquinolone group has mushroomed around the world,” says Dr. Bartlett, who co-wrote the editorial accompanying the articles by Drs. McDonald and Loo and their colleagues in the Dec. 8, 2005 issue of the New England Journal. Combined with the rise of a strain resistant to fluoroquinolones, ubiquitous use of FQs may account for an increase in the number of cases of CDAD, he says. “If you calculate the number of people with CDAD relative to the number who took fluoroquinolones, that is not a big impressive number,” he says. “It is not nearly as impressive as with clindamycin in the 1970s. But we make up for it in the number at risk—the volume of patients getting these drugs.”

Support for this assertion comes from the Pittsburgh data. In that outbreak the odds ratio for getting CDAD was higher for clindamycin (4.8) and ceftriaxone (5.4) than for levofloxacin (2.0). But the amount of disease attributable to levofloxacin (30.8 percent) was much higher than for clindamycin (10 percent) or ceftriaxone (6.7 percent). “It is the fluoroquinolones that are driving these outbreaks,” Dr. McDonald concludes.

More specifically, he points to the “respiratory FQs,” which are indicated for treatment of community-acquired pneumonia. Levofloxacin was introduced in 1997, moxifloxacin and gatifloxacin between 2000 and 2001. These agents became popular, especially in older people, who are already at increased risk for CDAD. What Dr. McDonald calls “striking evidence” for the idea that epidemics of CDAD could be driven by use of fluoroquinolones for respiratory infections comes from Canadian data showing high winter peaks of epidemic CDAD. That pattern coincides with the pneumonia season but not with the typical seasonality of diarrheal diseases, most of which have summer peaks.

While fluoroquinolone resistance may account for the greater incidence of the epidemic strain, increased toxigenic potential may cause its increased severity. Research on this point was reported by Michel Warny, MD, PhD, project leader, C. difficile Vaccine Program, and director of bacterial process development at Acambis in Cambridge, Mass. (Warny M, et al. Lancet. 2005;366:1079–1084). When Dr. Warny read and heard early reports about the epidemic strain in the U.S., he wondered if it was responsible for the outbreak of severe disease in Sherbrooke and whether it produces more toxin than previous common strains, possibly due to the deletion (tcdC) in what is thought to be a negative regulator gene of toxin expression. Using an in vitro system, he showed that the epidemic strain produces 16 times more toxin A and 23 times more toxin B than strains without the tcdC deletion. (Toxins A and B inactivate proteins that regulate the actin cytoskeleton.) “We do not know whether this in vitro feature is responsible for the increase in disease severity,” Dr. Warny says, “but it is an intriguing possibility given the principal role of toxins A and B as virulence factors.”

Dr. Gerding sounds a note of caution about these data. “In vitro toxin production does not correlate very well with virulence in C. difficile,” he says. “So we have to be a bit skeptical. Dr. Warny has clearly made a nice observation. But we are still missing a step. There may be more to virulence than high toxin production.”

The analysis Dr. Miller reported on behalf of a subcommittee of the Canadian Hospital Epidemiology Committee and the Public Health Agency of Canada is consistent with Dr. Warny’s data. CDAD patients infected with a strain carrying the gene for binary toxin were twice as likely to have a severe outcome (including death) as those without the gene for binary toxin. Dr. Miller notes that 93 percent of the isolates that were positive for the binary toxin gene were also positive for the deletion in the putative regulator gene. “We don’t understand whether binary toxin confers virulence or the deletion in the regulator gene or both,” he says. “They are so tightly correlated it is hard to separate their contributions.”

Two other important observations came out of this analysis. First, the incidence of CDAD across hospitals covers a broad range. “Our current hypothesis,” Dr. Miller says, “is that this figure is influenced by the type of antibiotic used and methods of infection control, as well as the presence of the epidemic strain in your hospital.” Hospitals that have three- and four-bedded rooms and cutbacks in nursing and housekeeping are “ripe for an epidemic,” in Dr. Miller’s view. Patient population is also critical. Dr. Loo and colleagues found that “probably the most significant factor in whether a patient gets disease or dies is age,” Dr. Miller points out.

Second, the epidemiology subcommittee reported the in vitro susceptibility of the hypervirulent strain to metronidazole and vancomycin, drugs that are used to treat CDAD. Contrary to anecdotal reports, they found no resistance to metronidazole and no resistance to vancomycin—yet.

With the increased virulence of the epidemic C. difficile strain, early and accurate diagnosis becomes critical. Currently, most laboratories in the U.S. test for toxins A and B using EIAs. While rapid, these assays are not very sensitive. According to Dr. Peterson, many investigators have found EIA tests to have sensitivity as low as 53 percent and suboptimal specificity—down to 75 percent. In 1995 the Society of Healthcare Epidemiologists of America published a guideline that included recommendations for diagnostic testing using toxigenic culture to optimize sensitivity and stool cytotoxin testing for enhanced specificity. Unfortunately, these tests take 24 to 48 hours and are not widely used. “Very few hospitals use either stool culture or cell cytotoxin assay [CTA],” Dr. Gerding says. He estimates a loss of sensitivity of 20 percent to 30 percent with EIA compared with culture or CTA.

“In our clinical laboratory we were using EIA, which gives a result in hours or the same day at least,” Dr. Gerding says. “We have now added culture to that.” However, the turnaround time for culture is 48 hours. “For sick patients we want a faster result,” he says. “I recommend doing cell cytotoxin assay and, if possible, cultures. If you want to use a rapid EIA,” Dr. Gerding suggests, “back it up with culture. Most laboratory people say that you can’t put that many resources into C. difficile testing,” he concedes. “But this is critical. I think laboratories have to start thinking about being more aggressive in this outbreak situation.”

Karen C. Carroll, MD, professor of pathology at Johns Hopkins University School of Medicine and director of the Division of Medical Microbiology at Johns Hopkins Hospitals, has put considerable thought into this dilemma. “Even before we had the epidemic strain, one of our problems was poor performance by our EIA for toxin A and B detection,” Dr. Carroll says. “We learned about this largely from physicians who took the time to call and basically state that they were seeing cases of fulminant colitis with tests coming back negative or indeterminate. Some had seen several negative results followed by one positive.”

After a series of those calls she went back and evaluated the performance of the EIA they were using. “Much to our surprise the level of sensitivity had fallen off,” she says. “I am still not sure why.” She evaluated another EIA, whose performance was similar—less than 80 percent sensitivity and inadequate specificity. Dr. Carroll finds that, in general, EIAs for toxin A/B fall short in terms of sensitivity (<90 percent). “I thought, we can’t go along with a test that is insensitive in the setting of a more virulent strain,” Dr. Carroll says. Moreover, single-use cards are not optimal for high-volume settings.

She also identified a collateral problem: Because clinicians did not trust laboratory results, they were sending three to six samples per week from patients with a high index of suspicion for CDAD. As a result, she says, “We were performing way more tests than we needed to.”

At that point Dr. Carroll abandoned EIA and went back to CTA for all samples. “Cytotoxin testing has greater than 90 percent sensitivity, but it has its own set of problems,” Dr. Carroll notes. It takes as long as 48 to 72 hours to confirm a positive result, which is burdensome to the laboratory. “So it did not satisfy our need for a rapid turnaround time.”

It was then that Dr. Carroll learned about a new EIA antigen test for glutamate dehydrogenase (Cdiff Chek, TechLab/Wampole), which identifies C. difficile but does not differentiate the epidemic strain or toxin-positive from toxin-negative strains. She found that this EIA performed quite well, with a high negative predictive value (a negative test means it is highly unlikely that the patient has C. difficile). Any test that is positive is confirmed by CTA. “That allowed us to do several things,” Dr. Carroll says. “We could eliminate negatives on the same day so the clinician could focus on other causes of the patient’s diarrhea. And it allowed us to set up CTA with neutralization.” CTA is read for cytopathic effect; a positive result must be confirmed as being due to C. difficile cytotoxins by including specific anti-toxin antibody in the culture. Once negatives have been screened out by the antigen test, CTA plus and minus neutralizing antibody can be set up in parallel, rather than in sequence.

This two-step algorithm was established about 1.5 years ago (Ticehurst JR, et al. J Clin Microbiol. 2006;44:1145–1149). “Everyone liked it,” Dr. Carroll says. Laboratory technologists and technicians liked the reduced work. Clinicians appreciated the increased sensitivity. As a result, she says, “We have been able to reduce the number of samples coming into the laboratory.” Another EIA that measures glutamate dehydrogenase (“common antigen”), along with toxin A, is the Biosite/Somagen Triage assay. “The Triage assay is very widely used,” says Dr. Miller. “In fact, this rapid 15-minute kit is used by many clinical study sites to enroll CDAD patients rapidly, once diarrhea is detected.” One group of investigators found that a positive Triage result for common antigen indicated sensitivity and specificity of 100 percent and 82.7 percent, respectively, for a negative predictive value of 100 percent, which, the authors wrote, “should prove useful in avoiding unnecessary use of contact precautions and antibiotic treatment for C. difficile-negative patients” (Landry ML, et al. J Clin Microbiol. 2001;39:1855–1858). However, they cautioned that CTA should be used to distinguish true cytotoxin-positive patients from C. difficile carriers.

Perhaps the final answer to the testing dilemma will be a molecular assay. At Evanston Northwestern Healthcare, Suzanne Paule, BS, and Dr. Peterson, working with Karen Kaul, MD, PhD, have been developing a real-time PCR assay for the toxin B gene. Interference was expected from inhibitors in stool. “We spent a lot of time talking about the extraction [from stool], but we have had fewer issues than we expected,” says Dr. Kaul, director of molecular diagnostics at Evanston Hospital and professor of pathology and urology at the Northwestern University Feinberg School of Medicine. For extraction she uses presteps and then the MagnaPure.

In an evaluation of the real-time PCR assay, Drs. Kaul and Peterson and their colleagues analyzed 429 stool specimens and found, compared with culture, 100 percent sensitivity and 98 percent specificity. EIA had 65 percent sensitivity and 92 percent specificity. Accurate determination of carrier status (positive PCR test but symptoms not due to C. difficile) was not possible by retrospective chart review, Dr. Kaul says. In some cases stools being analyzed were quite solid, that is, physicians were overordering. “We are doing our final prospective evaluation of our test,” Dr. Peterson says. The evaluation includes the PCR assay in parallel with EIA and CTA. “We are going to talk to patients to verify that they meet clinical criteria,” Dr. Peterson emphasizes. “I suspect if it works out we will convert this summer.”

Many hospitals will continue to use EIAs for toxins A/B. At the Reading (Pa.) Hospital and Medical Center, Sharon Strauss MS, SM (NRM), microbiology supervisor, had been using the CTA on stool filtrate. In spring 2005, however, pressure arose from physicians to get results more quickly because of an increase in the number of cases and increasing severity. “They needed to get patients into and out of isolation more quickly,” Strauss says. She evaluated the Meridian Immunocard and felt it compared favorably with CTA. Among 94 cases, there were only two discrepancies—one was probably a weak positive and the lab was unable to resolve the other. “Any rapid assay will be slightly less sensitive because CTA is the gold standard,” Strauss notes. They run the EIA twice each day.

Mary Beth Young, MT, SM (ASCP), medical technologist at St. Mary’s Hospital, Rogers, Ark., says the St. Mary’s laboratory did rapid assay for toxin A only. “When Remel came out with a toxin A/B rapid kit [Xpect] we switched to it,” she says. “That is one of the most requested tests that we do. It is a great kit.”

Controlling outbreaks of CDAD requires multiple steps. “The cause is multifactorial, so our approach has to be multifaceted,” Dr. Loo says. When an outbreak occurs on a ward, health care workers must shift from hand hygiene with alcohol gels to washing with soap and water, because alcohol-based products do not kill C. difficile spores. To decrease the C. difficile spores in the environment, diluted bleach must be used to disinfect rooms that house C. difficile patients. At McGill University Health Centre, Dr. Loo says, “We had the luxury of having a closed unit, so we minimized the number of patients per room, thereby decreasing the number of patients exposed should a patient in a multi-bed room develop CDAD.” When a patient develops CDAD, the patient is placed on isolation and moved to a private room. However, she notes, it is not always possible to immediately move a patient with diarrhea to a private room.

To strengthen infection control, Dr. Loo says, the Quebec Ministry of Health “stepped up to the plate” and gave hospitals CAN$20 million to fight C. difficile. That made it possible to purchase more dedicated equipment and hire more infection control practitioners and housekeeping staff. In addition, empiric treatment has been instituted: Patients are started on metronidazole after a first diarrheal episode pending test results.

With these measures, the incidence of CDAD in Quebec has been cut by half but is still twice the original baseline. In Dr. Loo’s view, antibiotic restriction could get it back to baseline. Thus far, they have attempted to control antibiotic use through resident education. “Our next step is to think of alternatives to fluoroquinolones,” she says.

Dr. McDonald, too, emphasizes appropriate hand hygiene—conventional washing and gloves—when caring for patients with diarrhea in an outbreak situation. He calls contaminated hands “the most important factor in transmission.” Also critical is the use of bleach on bed rails, commodes, bed pans, and other equipment that can become contaminated. “Quaternary ammonium compounds used in hospitals do not kill spores,” he stresses.

Over the long term some patients at high risk for CDAD may be protected by a vaccine. Dr. Warny and his colleagues at Acambis have completed a phase 1 study, in healthy adults between ages 18 and 65, of a candidate vaccine composed of a mix of inactivated toxins A and B. “The vaccine was highly immunogenic and well tolerated,” he says. They are now testing it for safety and immunogenicity in healthy people older than 65 years. Two indications are being considered: prevention of C. difficile relapses, which occur in about 20 percent of cases, and prevention of primary infection in patients at highest risk. Preliminary evidence that this vaccine may stop relapses came from compassionate use in three patients with multiple episodes of recurrent CDAD who had required nearly continuous treatment with oral vancomycin for seven to 22 months. After vaccination all three subjects discontinued treatment with vancomycin without further recurrence (Sougioultzis S, et al. Gastroenterology. 2005;128:764–770).

Overwhelming evidence proves the critical role of antibiotic therapy as a predisposing factor in CDAD—certain antimicrobials open up a niche for C. difficile in the normal colonic flora. But could some cases arise without prior antibiotic treatment? An investigation in Pennsylvania sought CDAD cases among people thought to be at low risk—peripartum women and people who had not been in the hospital recently. Among a total of 33 cases (10 peripartum) from four states (New Hampshire, New Jersey, Ohio, and Pennsylvania), eight reported no exposure to antimicrobial agents within three months before CDAD onset (MMWR. 2005;54:1201–1205). The authors raised the possibility that “[T]hese cases of severe CDAD disease in populations previously thought to be at low risk might further reflect the changing epidemiology of CDAD.”

A separate publication reported that CDAD had increased greatly in a community database in the United Kingdom and that the primary risk factor was use of proton pump inhibitors (Dial S, et al. JAMA. 2005;294:2989–2995).

However, experts in C. difficile disease view these ideas with caution. “For 25 years we have been looking hard for anything that causes this disease other than antibiotics,” Dr. Bartlett says. “In that time we have seven case reports of people getting C. difficile disease without taking an antibiotic. That is how rare it is. This is one of the tightest associations known. So I am not convinced by that data.”

Dr. Miller, too, is skeptical about the possibility of getting CDAD without prior exposure to antimicrobials. “In 20 years of working with C. difficile,” he says, “I could count on one hand the number of patients who have developed CDAD without an antibiotic in recent time.” In the analysis of data from the Montreal outbreak they looked for a link to proton pump inhibitors. “We did not see PPI use as a significant factor for developing CDAD,” he says. “I think that many of us believe that PPI use is not a major risk factor and probably will not by itself give you CDAD.” Investigators in Toronto have just finished an outpatient study on PPI use and C. difficile infection, Dr. Miller says. “Their data will be out soon.”

Proving that CDAD can be acquired without prior antibiotic exposure “would be a clear paradigm shift,” Dr. Gerding says. “Virtually all of C. difficile disease is acquired in hospitals in persons treated with an antibiotic.” Outside hospitals only a small percentage of people have C. difficile in their stools—one percent to two percent. Upon admission to the hospital, the rate of C. difficile in stool increases linearly by seven percent to eight percent per week, he says.

He notes that almost everyone who got severe CDAD in recent outbreaks was critically ill or elderly. “Clearly the over-65 population is most susceptible,” Dr. Gerding says. “Now, based on the MMWR report, everyone is alarmed about pregnant women and children getting it. But that is a tiny handful. And it is not even clear that it is different from what we have seen in the past.” The 33 cases in the MMWR article were gathered over 28 months, Dr. Gerding notes. “And only two C. difficile isolates were identified out of all those patients. So it is not clear what they were due to.”

Efforts to understand the ongoing outbreak of CDAD include tracking its spread across Canada. In their just-reported analysis, Dr. Miller and colleagues found three high-incidence provinces—Quebec, Ontario, and British Columbia. Now they are generating a dynamic picture of what is happening with the epidemic strain. “What drives movement of this strain from one part of the country to another?” he asks. “We will see whether there is increased fluoroquinolone use among hospitals where the incidence of the hypervirulent strain increases.”

Dr. McDonald believes that fluoroquinolone restriction needs to be tested as a way of containing the epidemic strain. “There are some data from England and the Netherlands where they banned all fluoroquinolones and saw outbreaks caused by the epidemic strain promptly cease,” he says. In one hospital, when fluoroquinolones were reintroduced, the outbreak strain returned. “These data are not surprising given past experiences with clindamycin restriction,” Dr. McDonald says. “I have not found many U.S. hospitals with the gumption to ban all fluoroquinolones, but this is where we could all benefit from a little courage.”