In the United States, a recent development has been the promotion of publicly reported performance measures for hospitalized patients, intended to assure a minimum standard of care. The latest IDSA/ATS guidelines have addressed these performance measures, endorsing some, but rejecting others, particularly the standard related to the timing of antibiotic administration. For patients with CAP, Medicare (Center for Medicare Services) has developed a set of "core measures" that are collected for every hospital and reported on the Hospital Compare Web site, with the implication that hospitals with the best performance are providing the highest level of care ( Table 3 ). The core measures include timely administration of antibiotics, measurement of oxygenation, selection of the correct antibiotics in the hospital and in the ICU, collection of blood cultures in seriously ill patients prior to antibiotic administration, provision of smoking cessation advice, and provision of pneumococcal and influenza vaccines to at-risk individuals. Because of either a high level of success with some measures (assessment of oxygenation) or problems (antibiotic timing), the performance measures have been periodically revised.
This approach has led to a variety of adverse, unintended consequences, best illustrated by the previous standard of requiring all patients to receive their first dose of antibiotics within 4 hours of arrival to the hospital, a standard that was modified to extend to 6 hours. Although retrospectively collected data have shown a lower mortality for patients given therapy within 4 hours, compared with those given therapy later, the explanation may not be timely administration alone. It is possible that delay in therapy may be a surrogate marker for patients with diagnostic uncertainty and of patients with impaired host responses who have indistinct clinical presentations and thus are not easily recognized as having pneumonia. In addition, when hospitals have increased the number of patients getting early antibiotics, in an effort to comply with core measures, there are data to suggest that there are more patients who are given timely therapy who do not actually have pneumonia, than in the period before such an effort at timely therapy was made. In addition, some of the patients given this timely therapy did not have pneumonia, yet developed antibiotic complications such as antibiotic-associated colitis.
Although the standard for pneumococcal vaccination seems appropriate, when the vaccine has been used in children it has led to unintended consequences, some useful, and some not. Initially, widespread use of the seven-valent conjugate vaccine in children led to "herd immunity" in the older caregivers of vaccinated children, with a decline in the frequency of infection by vaccinated strains. However, more recently, in the children who have been vaccinated, there has been an emergence of "replacement strains," with infection being caused by pneumococcal strains that are not included in the vaccine, some of which have led to necrotizing pneumonia.
Guidelines outside the United States, including the recently published European guidelines, have not formally included performance measures, and the value of these standards needs careful consideration. The 2007 US guidelines endorse four performance measures, which do not include all of the Medicare core measures. These are use of empirical therapy that is consistent with guideline recommendations because these therapies have been shown to reduce cost, length of stay, and mortality; giving the first dose of therapy in the emergency department, after establishing the diagnosis of CAP (but there is no endorsement of a specific time period for administration); collecting data about CAP mortality and the need for ICU transfer after admission to a medical ward; determining how many at-risk patients receive pneumococcal and influenza vaccine.
In the end, it may be necessary for all guidelines to consider performance measures because it has been well demonstrated that the mere presence of guidelines leads to little change in behavior. If guidelines do improve outcomes, then there should be some monitoring of whether guidelines are used and of the outcomes and benefits that follow from their use. However, in order for this to occur, we will need to better understand how to implement guideline recommendations. One US study showed that the best implementation and outcomes occurred when the presence of a guideline was supplemented by a clinical champion who prospectively intervened to make sure that recommendations were being followed. Although this approach worked in one hospital it is clear that each health care system will need to approach implementation in a unique way, and this may be the greatest justification for developing guidelines unique to each country so that there will be maximum "buy in" when efforts are made to implement these guidelines to improve patient outcomes.