| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
Dr. Waterer is Senior Lecturer in Medicine, University of Western Australia, Royal Perth Hospital.
Correspondence to: Grant Waterer, MD, FCCP, University of Western Australia, Royal Perth Hospital, GPO Box X2213, Perth 6847, Western Australia; e-mail: waterer{at}cyllene.uwa.edu.au
There is no doubt that the best diagnostic strategy in patients with suspected ventilator-associated pneumonia (VAP) remains contentious. The central problem is the difficulty in striking a balance between avoiding a delay in starting antibiotics when they are required and reducing inappropriate use of broad-spectrum antibiotics. Failure to achieve the first goal leads to excess mortality,1 2 3 while failure to achieve the second is a major factor in the seemingly inexorable increase in multiantibiotic-resistant pathogens with their attendant morbidity, mortality, and economic costs.
After numerous "test A" vs "test B" studies attempting to address the issue of the "best" diagnostic strategy, it became apparent that a clinical, microbiological, or pathologic diagnostic "gold standard" for the diagnosis of VAP was not achievable. The acknowledgment that no "gold standard" was going to be found led to the plea to shift the focus of research from comparative assessments of diagnostic techniques to outcome-based studies.4 5 Although two studies have made important starts in this direction,6 7 their apparent opposite findings with respect to the value of invasive diagnostic techniques has lead to much debate about their relative merits but no progress toward any consensus.
One of the significant limitations of clinical studies of VAP is that patients are enrolled if they meet a set of criteria, but once entered there is no stratification into separate risk groups based on clinical likelihood of VAP. The study by Sirvent and colleagues in this issue of CHEST (see page 518) demonstrates how nonspecific even relatively strict clinical criteria are for the diagnosis of VAP (at least matched against a microbiological standard). In this study, patients had to be receiving mechanical ventilation for > 72 h, and have a new or changing pulmonary infiltrate, fever, leukocytosis, and purulent tracheal secretions. At face value, these patients would appear to be highly likely to have VAP, yet 21% were determined to have an alternative diagnosis. However, despite all patients being grouped as possible VAP, it is unlikely that the treating clinicians would have estimated the same risk of VAP in all patients. The clinical perception of the likelihood of VAP and the manner in which the test result will influence management plays a significant role in what performance characteristics are desirable in any diagnostic test.
There are at least three different clinical scenarios physicians encounter in the setting of suspected VAP that will influence the choice of diagnostic strategy. Firstly, the treating physician may decide to start antibiotic therapy regardless of the result of any diagnostic test. In this case, there is no opportunity for a reduction in empiric antibiotic use, although a test result may lead to a change in antibiotics or influence the duration of antibiotic therapy. Although this strategy significantly reduces the potential benefit of testing (ie, there will be no reduction in inappropriate antibiotic use), it may be quite valid in the appropriate clinical setting. Ruiz and colleagues7 employed this strategy in their patients with suspected VAP and found no advantage of invasive techniques over noninvasive techniques.
In the second clinical scenario, VAP is suspected but antibiotic therapy would be withheld if a diagnostic test result was negative. Although this approach optimizes the potential for a diagnostic test to influence management, it has come under much criticism because of the potential for delaying the diagnosis of VAP. Despite these valid concerns Fagon and colleagues6 demonstrated in the patients they studied that withholding antibiotics when the Gram stain of BAL fluid was negative reduced antibiotic consumption without increasing mortality. Therefore, while delayed antibiotic therapy may increase the risk of mortality when assessed over all cases of VAP, it may be quite possible to define a group of patients (either clinically or in combination with diagnostic studies) where a delay in initiating antibiotic therapy because of a false-negative test result will not adversely affect outcome.
In the final scenario, the treating clinician may have a low suspicion of VAP and would withhold antibiotics unless a diagnostic test result was positive. In their ICUs, Sirvent and colleagues would commence antibiotic therapy if 
2% intracellular organisms (ICOs) were detected on mini-protected BAL in an antibiotic-naïve patient. However, based on the findings of their study, the detection of ICOs does not provide any better information than a Gram stain, making it difficult to justify the additional expense.
In each of the scenarios described, the performance characteristic of the optimal diagnostic test will be different. In the first scenario, screening tests such as ICO detection or Gram stains have little value, as they will not influence management. The focus of the diagnostic strategy in this setting should be on optimizing its impact on the duration (or possible type of) antibiotic therapy. An example of this is the use of the cardiopulmonary infections score by Singh and colleagues8 to determine whether antibiotic therapy could safely be continued after 72 h. In the second scenario, the concern is false-negative results, so any test will have to be highly sensitive, possibly at the expense of specificity. In the third scenario, false-positive results will lead to overuse of antibiotics, so the optimal test will be weighted toward specificity.
Although other scenarios may well be defined, the importance for clinical studies is to recognize that it is the potential for a test result to impact on antibiotic use that affects the desirable performance characteristics of that test. If we are to make significant progress toward reducing antibiotic consumption, it is also important to try and accurately define subgroups of patients with suspected VAP in whom the second and third management scenarios described above can be employed.
Finally, it is extremely unlikely that any test or diagnostic strategy will ever approach 100% sensitivity and specificity. Any diagnostic algorithm will therefore strike a balance between minimizing underdiagnosis of VAP and minimizing overuse of antibiotics. Accurate measurement of the impact of a diagnostic strategy on these end points is critical. Researchers should also recognize that where the balance between these two outcomes is drawn (how much excess morality or morbidity due to cases of VAP missed for how much reduction in the adverse consequences of overuse of antibiotics) may vary between countries, as this decision is driven by societal values, health politics, and economics.
In conclusion, there is unlikely to be a single best diagnostic strategy for all cases of suspected VAP. Critical to generalizing the findings of any study is defining in advance how the diagnostic strategy employed will be used to impact on the timing, choice, and duration of antibiotic therapy. Clinical stratification of patients with suspected VAP into different risk (and hence management) groups may help optimize the beneficial impact of any investigation on outcome. While the effect of a diagnostic strategy on patient mortality, morbidity, and antibiotic consumption can be measured in a relatively short time span, definitive answers will have to await assessment of longer-term data on the impact (if any) on antibiotic resistance and its complications.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
