There’s this light on my way to work that is just a royal pain. It’s set up so that you have to wait for the arrow to make a left hand turn. The intersection is busy, especially in the mornings, and the arrow only stays lit for about 13 seconds. So you end up waiting five minutes or more – through several light cycles – to make the turn.
OR … you can go straight through the intersection, turn left into McDonald’s parking lot, pull out of the parking lot, come back to the intersection from the other direction, and make a right turn, saving yourself 4 minutes and 30 seconds.
Now mind you that drivers who choose the latter route are, in effect, going through a red turn arrow – they’re just taking a bunch of extra steps to make sure that they are complying with all of the traffic laws in the process.

You’re probably wondering what a traffic light has to do with bloodstream infections. I’ll get to that later.

This month, Consumer Reports published a well-written article about reducing hospital infections, and a lot of the take-home messages are good ones. The Consumer Reports article focuses on blood stream infections – also known as “septicemia“. Consumer Reports compared central line infection data for intensive care units at 926 hospitals in 43 states. Hospitals voluntarily submit such information to the Leapfrog Group, a nonprofit organization based in Washington, D.C. and Consumer Reports obtained the data from Leapfrog.

As many people realize, septicemia and sepsis can lead to significant mortality in patients. Approximately 20–35% of patients with severe sepsis and 40–60% of patients with septic shock die within 30 days. Anything that we can do to prevent bloodstream infections will be a net positive for patient care.

So it was interesting to read the data Consumer Reports collected regarding central line-related bloodstream infections. In every state, hospitals significantly decreased the number of central line infections that occurred. In fact, many hospitals – several with more than 6,000 central line days – reported ZERO central line-related blood infections. You read that right. ZERO. Zilch. Nada. Absolutely no incidents of central line-related bloodstream infections.

The prevention in central line-related infections is credited to a simple five step checklist that was developed by Peter Pronovost, a Johns Hopkins critical care specialist. He felt that public disclosure of infection rates was a powerful motivator for hospitals to reduce the incidence of infections.

I agree, to a point, but there is a bigger motivator out there, though. Cold hard cash.

Under Section 5001(c) of the Deficit Reduction Act, the Centers for Medicare and Medicaid Services was required to select diagnosis codes that “have a high cost or high volume”, results in higher payment, and “could reasonably be prevented using evidence-based guidelines.” Bloodstream infections related to catheters was chosen as one of these codes and eventually became known as a “never event” – at least alluding to the notion that such infections should “never” happen and making a firm statement that the government would “never” pay for care related to such infections. In law, the concept of incurring liability for the occurrence of an event, regardless of whether that event is within one’s control is called strict liability. Here are come comments I previously made about strict liability in medicine.
Faced with public scrutiny and the possibility of being held liable for providing significant amounts of uncompensated care to sepsis patients, hospitals needed to make changes … and they did.

So first I’d like to start by congratulating the hospitals in Pennsylvania that made the Consumer Reports list for ZERO central line-related bloodstream infections.
At the top of the list was UPMC Presbyterian – Shadyside. Shadyside was not only tops in the state, it was tops in the NATION. Shadyside had 13,596 patient “central line days” without a single central line-related infection. Amazing.
Also included in Pennsylvania’s list were UPMC St. Margaret in Pittsburgh with 2,902 infection-free central line days, UPMC Magee Women’s Hospital in Pittsburgh with 1,600 infection-free central line days, and Southwest Regional Medical Center in Waynesburg with 1,040 infection-free central line days.

Congratulations to these hospitals on jobs well done.

You’re probably wondering why I chose to look at the hospitals in Pennsylvania, aren’t you?

As part of the public shame er, um, disclosure efforts required under Pennsylvania law, Pennsylvania created a web site to compare various costs of treatment and efficiency of health care for multiple different medical problems. Pennsylvania collects information on more than 4.5 million patient visits each year and then summarizes that information on its Health Care Cost Containment Council web site (which it calls “PHC4″).
It just so happens that one of the metrics on the PHC4 web site is “septicemia” – those same “blood infections” that Consumer Reports wrote about.

Now if all four hospitals dropped their cathether-related blood infections to ZERO, then the incidence of blood infections should also decrease at least a little, right?

Let’s look at UPMC Shadyside. Even though the number of catheter-related blood infections was ZERO, the cases of septicemia increased each year between 2002 and 2008, and they increased a lot. As in 145 cases in 2002 up to 881 cases in 2008. The costs to treat those cases also increased – from $30,000 to more than $69,000 per event. AND their “outlier” numbers for prolonged length of hospital stay in patients with sepsis were worse than expected between 2006 and 2008.

UPMC St. Margaret’s data also showed an upward trend, from 152 cases of septicemia in 2002 to 250 cases of septicemia in 2006 and then down to 209 cases by 2008. Costs also more than doubled during that time period, reaching $37,228 per case by 2008.

Southwest Regional was the only hospital that had a downward trend of septicemia cases, but even that data was haphazard. 32 cases of septicemia in 2002, 40 cases in 2004, 14 cases in 2006, and 23 cases in 2008. The costs for treating septicemia at Southwest Regional also doubled, but in 2008, its charges were $16,253 – less than one quarter of UPMC Shadyside charges for treatment of the same medical problem.

Magee-Women’s Hospital also had strange data. The number of septicemia cases it reported remained between 5 and 9 per year from 2002 to 2005. Suddenly in 2006, the number of cases at Magee-Women’s jumped to 28 and remained between 23 and 28 per year from 2006 to 2008. Its costs increased by almost double from 2004 to 2008, reaching $41,288.

You’re probably thinking that other variables can affect this data, and I’d agree with you. Perhaps more people in Pennsylvania just happened to develop non-catheter related bloodstream infections during those years. Maybe all the other hospitals except for those above are getting contaminated central line kits delivered to them. Maybe some hospitals focus so much on preventing catheter associated bloodstream infections that they drop the ball in other areas. Who knows what other facts may explain the precipitous fall in catheter related bloodstream infections despite a significant increase in bloodstream infections as a whole. It just puts a question in my mind. Are things really getting better or are hospitals all over the country just telling us … and CMS … what we want to hear?

Think about it. For the sake of example, I’m going to use UPMC Shadyside because of their high volume of patients. Assume in 2008, that 10% of the patients with septicemia at UPMC Shadyside were Medicare patients with catheter-related bloodstream infections (this article from Great Britain cites catheter related bloodstream infections as 10%-20% of all hospital acquired infections in the UK, so I’m staying on the low side of the cited statistics). If all those infections were considered “never events,” Shadyside would have lost more than $6 million dollars in 2008 on the care of those patients. Every patient with a catheter-related bloodstream infection at Shadyside can translate into more than $69,000 in lost revenue for the hospital.

With reimbursements being cut and many hospitals bleeding red ink, you think that every hospital out there doesn’t have an incentive to selectively interpret bloodstream infection data?

Here are some examples of how that selective interpretation might occur.

All of the data that I could find relates to catheter associated bloodstream infections in the Intensive Care Unit. If a patient develops signs of an infection and is then moved out of the ICU before official culture results come back, does that patient get dropped as a data source? Don’t know. I couldn’t find any guidelines on what to do in that situation.

How is a “catheter associated bloodstream infection” even defined? There’s no universal definition. Even the CDC admits that “the rate of all catheter-related infections (including local infections and systemic infections) is difficult to determine. Although CRBSI ["catheter related blood stream infections"] is an ideal parameter because it represents the most serious form of catheter-related infection, the rate of such infection depends on how CRBSI is defined.”

We can use the definition from the National Nosocomial Infections Surveillance System requiring “presence of recognized pathogen” in blood cultures not “related to” infection at another site. What if the pathogen was not specified? Perhaps only gram positive cocci but subtyping not performed. Does that data get thrown out? What if the patient has a pimple at another site? Is that “related to” the blood stream infection? Does that data get thrown out? What if there is a bedsore anywhere on the patient’s body? No longer a catheter-related bloodstream infection?

Appendix A of this MMWR report (.pdf download) has other definitions. One definition requires that the same organism be cultured from the blood and the tip of the catheter that has been removed. What if the catheter tip wasn’t cultured? Another definition requires that two blood cultures at different times show the same organism. What if only one blood culture was done?

The definitions don’t say anything about antibiotics, either. If a patient receives antibiotics prior to blood cultures being drawn, it is likely that the antibiotics in the bloodstream will inhibit bacterial growth and will falsely decrease the numbers of positive blood cultures. If the patients get antibiotics through their central lines, how do you think that will affect the results of the cultures of the tips of the central lines? Is that reportable?

Leapfrog Group and the federal government make a big deal about paying for performance. “Tie payment to outcomes” the Leapfrog Group advocates. When you start tying payments to outcomes without a well-thought out plan on how to reliably measure the outcomes, you’re going to get exactly what you pay for. Garbage in, garbage out. Just like  drivers trying to avoid waiting five minutes to turn a corner when they’re late for work, hospitals have an incentive to avoid undesirable situations by taking advantage of loopholes in the rules and definitions.

The thing that bothers me most about data like this is that it tends to make people both complacent and angry.
People become complacent when they go to hospitals with “zero” catheter related bloodstream infections. What a great place this must be! I’m safe here! Maybe that’s true, but maybe it isn’t true. How is their data interpreted?
People become angry when they’re affected by one of these highly-publicized negative outcomes.  Hospitals that still “allow” patients to develop such infections are viewed as negligent and get a bad reputation.

Does this mean that hospitals shouldn’t follow the Dr. Pronovost’s five step checklist? Absolutely not. But if those checklists work sooooo well, then why doesn’t the government just say “we’re not going to pay you if you don’t use the checklist”? Focus on the process, not the outcome. You’ll get everyone following the checklist overnight. Then you’ll see how effective it really is.

Nah. There’s more political capital in making the agencies look good and making the hospitals look bad.

What’s the point of this protracted post? There are a few.
1. You get what you pay for. If you pay for statistics showing a decrease in some measured outcome, you’ll get statistics showing a decrease in some measured outcome.
2. You don’t get what you don’t pay for. When you stop paying for an outcome, those providing the services might find a way to avoid the outcome, they might find a way to make it look like the outcome never happened, they might find a way to make someone else pay for the outcome, or they just might stop providing the services altogether.
3. The devil is in the details.

Now, what’s all this about CMS representatives marching in some parade … with an Emperor?

P.S. Did anyone see any government run hospitals in Consumer Reports’ list? I didn’t.