J-J-J-JCAHO!!!!

JCAHO B.IT.C.H.E.S. – QA measures and Community Acquired PNA (CAP)

THE QUESTION - A lot of “too-doo” has been made recently by the powers that be in our departmetn regarding antibiotics given within a 4 hour window for patients suspected of having community acquired pneumonia (CAP), with an expected 90% compliance rate. Being individuals inherently opposed to rigid guidelines, Stack and I wondered where and why these guidelines came from, and whether there will be any adverse outcomes from a seemingly arbitrary standard such as this (i.e., increasing drug resistance). Granted, as residents, armed with any such data, we are unlikely to take down the monolith of JCAHO and the rest of the bureaucracy, but, will the data give us a way to BITCH, EBM style?

THE ANSWER (in parts):

I. THE BACKGROUND : Community acquired pneumonia (CAP) is currently one of four diagnoses identified by the federal Centers for Medicare and Medicaid Services (CMS) to be used as measures of quality improvement (the others being ACS, heart failure and surgical wound infections). In case you don’t know, CMS is a big agency – they gave out over $480 billion (that’s $480,000,000,000) in funding related to medicare and medicaid in 2004. When our wonderful president, GW Bush signed the Medicare Prescription Drug Benefit in 2004, the bill also provided $400-450 million annually to help incentivize quality improvement indicators in these diagnoses. In order to get their share of the funding, hospitals that accept medicare must submit their performance data for 10 quality measures, 6 of which fall under the realm of CAP (see below, data available at hospitalcompare.hhs.gov):

Quality Indicator BIDMC Average (USA) Average (MA)
1. Assessed/Given Influenza Vaccine 21% 70% 75%
2. Assessed/Given Pneumococcal Vaccine 93% 67% 71%
3. Given Abx w/in 4hrs of arrival 91% 79% 80%
4. Oxygenation measured 100% 99% 100%
5. Given smoking cessation advice/counselling 94% 79% 80%
6. Given most appropriate initial Abx 86% 83% 85%
7. Blood Cx given prior to Abx 94% 90% 90%

As you can see, BIDMC performs fairly well compared to other hospitals. So why the emphasis on full compliance? Color me cynical, but it appears that the possibility of future “pay per performance” is a likely reason. Currently, this is being piloted in the “Premier’s Hospital Quiality Incentive Demonstration Project,” a program started among 268 Medicare recipient hospitals in 2003. In this pilot, hospitals in the top 10% of quality performance measures recieve a 2% bonus, and hospitals in the second decile receive a 1% bonus. And coming soon, the bottom 20% will also be penalized 1-2% per annum as well. Many commentators see this as a future direction of all Medicare-recipient hospitals.

II. THE DATA.
So… to the point of this discussion – quality indicator number 3 above – Antibiotic administration within 4 hours of arrival for patient admitted with a diagnosis of CAP. Where did this data come from? It is, in fact, based entirely on two retrospective studies.
The first, Meehan et al (JAMA 1997) looked at 14,069 medicare patients > 65 yo who were admitted for CAP. Excluded from this study were pts < 65 yo, HIV+, transplant recipients, recently hospitalized, chemotherapy within 2 months, transferred from other hospitals, and patients who either did not recieve antibiotics, or recieved them > 100 hrs after admission. The primary outcome was 30d mortality rates. This study showed a marginal, but statistically significant improvement in 30 day mortality when patients recieved antibiotics within 8 hours of arrival (OR 0.85; 95% CI 0.75-0.96). Interestingly, they also showed that patients who recieved antibiotics within 1 hour had a 20% increase in mortality rate at 30days. Hmmmm. Note, the analysis all tried to control for “how sick the patient was” by adjusting based on the Pneumonia Severity Index (PSI; a prospectively studied and validated severity scoring system),
The second study, Houck et. al. Arch Intern Med 2004 is very similar. They retroepectively studied the medical records of a random sample of 18, 204 medicare patients again > 65yo with community acquired pneumonia. The primary outcomes were severity adjusted mortality (both in hospital and within 30 days), readmission rates within 30 days, and lenght of stay. Unlike Meehan et al, who used 8 hrs as a cutoff, these used the now-institutionalized 4 hour window. They found (and please note again, mortality rates are severity adjusted by the PSI):

Abx < 4 hrs Abx > 4 hrs OR (95% CI)
In hospital mortality 6.8% 7.4% 0.85 (0.74-0.96)
30d mortality 11.6% 12.7% 0.85 (0.76-0.95)
Length of stay > 5d 42.1% 45.1% 0.90 (0.83-0.96)

Yeah… not too impressive, I know. An absolute decrease of 0.6 %(in house) or 1.1% (30d) mortality rate. Interestingly, they also found a “trend” that patietns who had recieved antibiotics within 24hrs prior to admission (super-duper early!!!) had a increased mortality rate (OR 1.18 (CI 0.97-1.45), p=0.10).
What to make of this data? Well, first of all, from a methodological standpoint, all of these are retrospective, medical review studies, with all of the inherent problems (unmeasured confounders). For example – maybe patients who have a delay in diagnosis of CAP, and therefore delayed antibiotic administration, are inherently sicker – eg, maybe patients with pneumonia diagnosed by CT scan (and therefore, presumably later in the course) had a negative CXR because they were hypovolemic, or they were too sick to go for an PA and lateral (and had a crappy, false negative single view AP), or the provider was worried enough to get a CT scan for another reason.
Secondly, other very similar studies with very similar methodologies show absolutely no influence of timing to antibioitcs on mortality or length of stay measures! (See for example, Dedier J et al, Arch Intern Med 2001). There are no prospective studies that have confirmed the importance of early antibiotics in CAP.
Finally, it is also noteworthy that the current JCAHO reccomendations (that we are held to) for early antibiotics are not only for patients > 65yo (the population that the, albeit limited, data has been studied in), but for all patients > 18 yo – patients who would have been excluded from these same studies.
III. Conclusions

Conclude whatever you want from the above. It seems to me that the present requirements for 4 hour antibiotics are, at best based on very sketchy, non-prospectively validated data for a specific patient population of Medicare patients (older than 65). At worst, they are artificial constructs being implemented more for the financial benefit of the hospitals than for the medical benefit of the patient. In fact, it is not too hard to imagine situations in which this could actually lead to a net harm to our patients – will the increased/less discriminant use of antibiotics lead to more antimicrobial drug resistance, or more adverse reactions to antibiotics? Will patients with possible pneumonia be triaged faster from the waiting room to meet these requirements than patients who are sicker but aren’t “lucky enough” to have diseases that are tied into the hospital reimbursement system? I don’t know… but these seem to be real risks, all the worse if “pay for performance” takes hold, or god forbid, expands.