Editor's note: Sanjay Saint, MD, MPH, is Professor of Medicine at the University of Michigan and the Ann Arbor VA Medical Center in Ann Arbor, Michigan. Dr. Saint's research has focused on reducing health care–associated infections, with a particular focus on preventing catheter-related urinary tract infections (UTIs). We asked him to speak with us about how research on UTI prevention provides broader lessons for patient safety.

Dr. Robert Wachter, Editor, AHRQ WebM&M: What got you interested in health care–associated infections in the first place?

Dr. Sanjay Saint: When I was a house officer at UCSF, I spent my last year as a chief resident at the VA. And it struck me that sometimes veterans who had been there for several weeks, when we asked the team for the reason of the initial admission, they had forgotten—because they spent most of their time dealing with complication after complication. Venous thromboembolism, ventilator-associated pneumonia, central line–related infection, and urinary catheter–related infection. I was interested in trying to figure out how to prevent some of these complications, did a literature search, and found that while there are a lot of data on the prevalence of these infections and their risk factors, there were much fewer data on what actually can be done to prevent them.

RW: You've done a lot of work in all of those areas, but you drilled in on urinary tract infections with particular interest. Why that?

SS: What interested me was that urinary catheter infections are so often overlooked, but it's such an integral part of patient care, and there really wasn't a lot of research in the area. So I thought that perhaps I could have an impact, even by doing simple types of studies such as trying to figure out how often physicians are aware that a urinary catheter is even present.

RW: I thought that was a very clever study. What gave you the idea to do that and what did you find?

SS: One of the interventions that I found most useful when I was an attending was to ask: "Can any of their lines and devices be removed?" Often I found that the residents said, "Well, does the patient even have a catheter?" And they would ask me, "Dr. Saint, do they have a catheter?" And I said, "You know, I'm not sure." And so we all went on the other side of the bed, lifted up the sheet, and saw that there was a catheter there. So, there was a sense anecdotally that the physicians weren't aware that the catheter was present. Data from other investigators revealed that between one third to half of the days that a patient had a urinary catheter, there would be no justified indication for it by CDC criteria. So I wanted to formally study whether or not physicians were aware that their patients had urinary catheters. We conducted a study of four different hospitals across the United States and found that, depending on the physician's training level, awareness varied. Medical students were unaware about 20% of the time. Interns and residents were unaware about 30% of the time, and attending physicians—who are medically and legally responsible for the care provided to their patients—were unaware almost 40% of the time that the catheter was present. And they were more likely to be unaware if the catheter was not indicated. One other finding I'll just bring up is that we also found that in about 30% of the hospitals, there was no documentation anywhere that the catheter existed. There was no physician order for one, even though one is required. It wasn't in the physician notes, and it wasn't in the nursing notes, but we knew it was there because we could see it coming out of the patient. So we call that phenomenon the "immaculate catheterization."

RW: How did your findings influence the next steps in your work?

SS: What we've tried to do is make urinary catheters an important part of what health care providers think about. But rather than relying on their memory and just educating them, because that seems to be a short-lived solution, we're approaching it in a systems way by studying urinary catheter reminders. We initially studied a urinary catheter reminder in the Seattle VA, which has state-of-the-art computerized physician order entry (CPOE) systems. After 72 hours, the intern was reminded that a patient had a urinary catheter. If they want it continued, they should just check off the indication, and if they did not want it continued, they could check off to have it removed. When we did the initial studies, the University of Michigan did not yet have CPOE (we do now), so we used a written reminder where a nurse went around after 48 hours and put a note on the medical record reminding the intern or resident that the patient had these urinary catheters. Again, that way they could have it removed, as no longer necessary. Others have also studied the urinary catheter reminder using a nurse-based reminder system, specifically in an ICU. All these studies have found a decrease in urinary catheterization rates. However, it's unclear whether or not this intervention, which had high face validity, also will lead to a decrease in urinary tract infection.

RW: As I recall, there's a study that just came out that looked at that connection and did not find a change in rates of infection.

SS: They found a significant reduction in urinary catheterization rates, but not a significant reduction in urinary tract infection rates. One of the other important aspects of urinary catheter use is not just the infections—we also know that patients don't like having a catheter through their urethra into their bladder. It leads to discomfort and pain—and in some cases, embarrassment. Importantly, it limits their activities of daily living. On the one hand, we're trying to get patients up out of bed to prevent deconditioning, decubitus ulcers, and pressure sores, but the urinary catheter is tethering them to the bed, preventing them from getting up and doing activities of daily living. Even if you leave aside the infectious complication, I would argue that the noninfectious complications are also important to consider in removing the catheter.

RW: There may be some trade-offs in terms of patients being forced out of bed and an increased fall risk. Do you worry about that at all?

SS: Yes, I think that is a potential concern. On the other hand, I also know of patients who have the catheter in and still try to get out of bed. Because often patients who need a catheter are cognitively impaired and may not feel the catheter, or in the middle of the night, they might not realize that the catheter is draining their bladder. They try to get out of bed and then they trip over the catheter tubing. So it's unclear which one is more likely to be a problem. The other approach to urinary collection is thinking about whether we can do things to avoid indwelling catheterization. For example, would intermittent catheterization, either done by a patient, a nurse, or a nursing aide, be useful? Other approaches would be using external-style catheters in men—for example, condom catheters, in which men who are able to urinate do so without having an indwelling device. (Unfortunately, an external device isn't really available for women because of logistical problems.) But for men, the condom catheter may be used, and patients in a randomized controlled trial found it to be more comfortable than the indwelling catheter, and in some patients it led to a lower risk of bacteriuria.

RW: One of the big policy interventions recently has been Medicare's decision, now being followed by private insurers, not to pay for certain preventable adverse events. Nosocomial urinary tract infection is on that list. Should it be on that list?

SS: I applaud Medicare's interest in preventing hospital-acquired complications in general, and hospital-acquired infections in particular. I suspect that, by no longer paying for catheter-related urinary tract infections, it will increase the interest that hospitals have in preventing these important complications. I think in the long run this will be a net plus. This type of policy intervention may also have problems of its own that need to be carefully monitored and evaluated, and then changes will need to be made. When it comes to catheter-related urinary tract infections, hospitals may decide to screen at the time of admission to see if patients have bacteriuria. We know that a small but significant percentage of patients will have bacteriuria, and they'll be completely asymptomatic. Now will that lead to an increased use of antimicrobials? Most people recommend that antimicrobial therapy for asymptomatic bacteriuria is unnecessary unless patients fit certain criteria—for example, they're going to get a renal transplant, or they're pregnant. But for the vast majority of patients, asymptomatic bacteriuria should not be treated with antibiotics.

The other concern about catheter-related urinary tract infection being on this list is that it differs from at least two of the others—leaving a foreign body in place and vascular catheter-related infection—in the amount of evidence supporting its preventability. Vascular catheter-related infection does seem to be largely preventable based on Pronovost's study from a couple of years ago showing a 66% and statistically significant reduction in infection in a large number of intensive care units in Michigan. Even though it wasn't a randomized controlled trial, this kind of quasi-experimental study makes people feel comfortable that a multi-modal, evidence-based approach can prevent vascular catheter-related infection. Such confidence in putting in place programs to prevent catheter-related urinary tract infections to my knowledge is not there.

RW: You spoke about the challenges and maybe some of the opportunities in the new policy of not paying for certain adverse events. The Joint Commission recently came out with 2009 safety goals, and preventing infection is a very big part of that. What do you think of the accreditation and regulatory approach to this problem?

SS: One of the important issues that we should focus on is making sure that the processes that have been shown in randomized trials or high-quality quasi-experimental studies to be linked to important clinical outcomes are in place at hospitals. For example, to prevent vascular catheter-related infections, the standard of care now would be that chlorhexidine be used rather than povidone iodine, and ideally it would be in the central venous catheter kits. That maximum sterile barriers be used rather than the limited type of barriers that we used in the past, and that the large drape be either in the kit or be part of the central venous catheter kit. For urinary catheter–related infections, urinary catheter reminders of some sort would be used—or at least a rationale for why they're not being used.

What would be useful from a regulatory point of view is to evaluate some of these process measures to see if the hospitals are doing them. And if not, there should be some reasonable justification given for why it's not happening. An example would be what we found in our national survey: that a small but nontrivial percentage of hospitals, roughly 10%, were not using either chlorhexidine to fight infection or maximum sterile barriers at the time of catheter insertion. That really shouldn't be acceptable given that there are good data to support at least chlorhexidine and reasonable data to support maximum sterile barriers. That would be one approach that could be used.

RW: What do you think the broader field of patient safety should be learning from your own research and the broader experience that people have had in the field of infection prevention?

SS: Well, I think that infection prevention in many ways is a paradigm for patient safety, at least for certain aspects of patient safety research. In the 1970s, when infection control really began to blossom, infection control professionals, the hospital epidemiologists, focused on coming up with standardized definitions using benchmark data, trying to understand what works and what doesn't through either randomized controlled trials or quasi-experimental studies, and then trying to intervene and seeing if it made a difference. For many patient safety problems, that type of applied epidemiological approach is appropriate: falls, venous thromboembolism, and contrast-induced nephropathy, for example. Issues where you have a numerator and a denominator and can come up with a rate lend themselves to an epidemiological and evidence-based approach. On the other hand are other patient safety problems that are much more rare and get a lot more attention: wrong-sided surgery, taking the wrong patient to have a procedure performed, and incompatibility of organs, for example. These types of things are so rare that approaching them epidemiologically may not be as appropriate as approaching them through human factors. The human factors approach and the epidemiological approach are complementary. In part, it depends upon what complication we're trying to tackle. What we have tried to do is focus primarily on complications that are common enough so that we could approach them in an epidemiological way.