Neutrophil Distribution Width Study Flawed/In Reply

To the Editor.-After reading ''Neutrophil Volume DistributionWidth: A New Automated Hematologic Parameter for Acute Infection'' by Chaves et al,1 I had one question: how did the article pass peer review? The study design does not support the authors' contention that the neutrophil volume distribution width (NDW) has utility for diagnosing acute bacterial infections. The authors compared NDWs between a group of patients with known acute bacterial infections and a group of patients whose automated complete blood count results were completely normal. Clinicians do not need a test to distinguish between patients with acute bacterial infection and patients with normal complete blood counts. In this study, the diagnostic process could not benefit from any hematologic test, given the authors' finding that ''All patients had clinical indications of acute infection.'' The study provided no evidence that the NDW would have clinical utility for diagnosing acute bacterial infection from a population that includes patients with acute stress response, hematologic disorders, chemotherapy and other drug toxicity effects, immunosuppression, infection, and so forth.

For a test to be useful, it must provide additional information to the clinician. The NDW does not do this. The disease group's prior probability of acute bacterial infection was 1.0. Therefore, additional tests for diagnosing infection cannot add information. Instead, the NDW test, if used as suggested, would cast doubt on the diagnosis for 21% of the patients who clinicians judged to have acute bacterial infection. The NDW has negative utility.

GREGORY TETRAULT, MD

Department of Veterans Affairs

Pathology & Laboratory Medicine Service

Memphis, TN 38104

1. Chaves F, Tierno B, Xu D. Neutrophil Volume Distribution Width. Arch Pathol Lab Med. 2006;130:378-380.

The author has no relevant financial interest in the products or companies described in this article.

In Reply.-We write in reply to the letter written by Dr Tetrault about our article.1 Although we welcome any constructive criticisms, we believe that the statements made in this letter were based on a poor understanding of our study design and the interpretation of our results. Therefore, we appreciate the opportunity to rebate such statements as the following.

1. ''The study design does not support the authors' contention that the neutrophil volume distribution width (NDW) has utility for diagnosing acute bacterial infections.''

The design of our study was a case-control study, classically the type of study initially used to test a new hypothesis. We needed patients who were definitely affected by the condition we were studying (infection), thus we selected our ''case'' group based on the gold standard test for diagnosing sepsis, the blood culture. On the other hand, to decrease the risk of introducing a confounding factor into the study, we wanted to assure that the control group did not include patients with conditions likely to also alter the NDW. Because such conditions would also likely alter the patient's complete blood count (CBC), we chose to consider a normal CBC as an inclusion criterion for the control group.

2. ''The authors compared NDWs between a group of patients with known acute bacterial infections and a group of patients whose automated complete blood count results were completely normal. Clinicians do not need a test to distinguish between patients with acute bacterial infection and patients with normal CBCs.''

In this statement, the letter's author implies that patients with normal CBCs could not be infected and all patients with acute bacterial infection have abnormal CBCs. Indeed, if this were true, clinicians would not need a test, such as the NDW, to distinguish between patients with acute bacterial infection and patients with normal CBCs. However, it is widely known that the sensitivity of the CBC in the diagnosis of infection is less than optimal, and patients with sepsis could have low, normal, or high CBCs. In fact, if one reviews our data, 32 of 70 patients with positive blood cultures had a CBC within the normal range, indistinguishable from the control group. In these patients, the NDW proved to be a statistically significant diagnostic discriminator between infected and control groups.

3. ''In this study, the diagnostic process could not benefit from any hematologic test, given the authors' finding that all patients had clinical indications of acute infection.''

Here, the letter's author assumes that having the clinical indications of acute infection is enough to diagnose this condition. If this is the case, there is no need to check an electrocardiogram or cardiac enzymes for those patients with clinical indications of myocardial infarction such as severe chest pain. In reality, the clinical signs and symptoms of infection overlap with those of several other acute medical conditions, so clinicians rely on laboratory tests like the CBC to confirm their clinical suspicion. The NDW is another potential laboratory parameter that hopefully will aid the clinician in making correct clinical decisions.

4. ''The study provided no evidence that the NDW would have clinical utility for diagnosing acute bacterial infection from a population that includes patients with acute stress response, hematologic disorders, chemotherapy and other drug toxicity effects, immunosuppression, infection, and so forth.''

Our study was a proof of concept study with 2 clear scientific questions: whether the NDW would differ at all between infected patients and controls, and whether any difference seen could be useful clinically. Our study was designed to answer those questions, and it did. Because this was a proof of concept study, we gathered data only on a small number of patients with bacterial infection, but we did acknowledge in our article the need for further studies, which are currently in progress. We also invite the author and others to pursue further investigations on the behavior of the NDW and the other volume, conductivity, side scatter (VCS) parameters in different clinical conditions, including those listed in the letter. Nevertheless, as with any laboratory test, its value will always be optimized when used in a broad clinical context, which would obviously include a patient's known history of any of the conditions listed here.

5. ''For a test to be useful, it must provide additional information to the clinician. The NDW does not do this.''

Although the clinical potential of the NDW in the differential diagnosis of the several conditions associated with leukocytosis still remains to be investigated, our study has already demonstrated at least 1 important use of the NDW: as an additional discriminator for septicemic patients who have otherwise unremarkable hematologic profiles.

6. ''The disease group's prior probability of acute bacterial infection was 1.0. Therefore, additional tests for diagnosing infection cannot add information.''

The prior probability could be an interfering factor in a cohort type of study, but in a retrospective case-control study, such as ours, it is impossible to calculate it because the cases and the controls are deliberately separated on the basis of the presence or absence of a given condition (in this case, acute bacterial infection).

7. ''Instead, the NDW test, if used as suggested, would cast doubt on the diagnosis for 21% of the patients who clinicians judged to have acute bacterial infection.''

Because the NDW does not have 100% sensitivity, it is true that there is a percentage of patients who would still have normal NDWs in spite of being infected. However, the same statement could be made for the white blood cell count and the neutrophil percentage, which in fact have even lower sensitivities. Because these are tests currently well accepted and widely used for the diagnosis of infection in spite of their limitations, it would be a mistake to disregard a new option such as the NDW, which, although not a perfect test, still performed better than the white blood cell count and the neutrophil percentage.

Finally, the basis of a laboratory test's usefulness and reliability is grounded in the hard science of statistics. Although a test with 100% sensitivity and 100% specificity is the ideal, in reality no tests perform so perfectly. Given the variability in testing and the potential for falsepositive or false-negative results, no diagnosis should ever be made solely on the basis of a single laboratory test. Laboratory tests are not foolproof and can present an incomplete picture when not put into proper clinical context.

FERNANDO CHAVES, MD

BETHANY TIERNO, MD

DONGSHENG XU, MD, PhD

Department of Pathology and Laboratory Medicine

Boston University Medical Center

Boston, MA 02118

1. Chaves F, Tierno B, Xu, D. Neutrophil volume distribution width: a new, automated hematologic parameter for acute infection. Arch Pathol Lab Med. 2006;130:378-380.

The authors have no relevant financial interest in the products or companies described in this article.

© 2006 College of American Pathologists Provided by ProQuest LLC. All Rights Reserved.