American Journal of Emergency Medicine 31 (2013) 687–689

Contents lists available at SciVerse ScienceDirect

American Journal of Emergency Medicine

j ourna l homepage: www.e lsev ie r .com/ locate /a jem

Original Contribution

Red cell distribution width as a predictor of mortality in acute pancreatitis

Kazım Şenol MD, Barış Saylam MD, Fırat Kocaay MD, Mesut Tez MD⁎

Department of Surgery, Ankara Numune Training and Research Hospital, Bahçelievler, Ankara, Turkey

⁎ Corresponding author. Tel.: +90 312 215 38 34; faxE-mail address: mesuttez@yahoo.com (M. Tez).

0735-6757/$ – see front matter © 2013 Elsevier Inc. Alhttp://dx.doi.org/10.1016/j.ajem.2012.12.015

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 2 November 2012Accepted 18 December 2012

Introduction: Acute pancreatitis (AP) is a common cause for hospitalization worldwide. Identification ofpatients at risk for mortality early in the course of AP is an important step in improving outcome. Red celldistribution width (RDW) is reflective of systemic inflammation. The objective of this study was to investigatethe association between RDW and mortality in patients with AP.

Methods: A total of 102 patients with AP were included. Demographic data, etiology of pancreatitis, organfailure, metabolic disorder, hospitalization time, and laboratory measures including RDWwere obtained fromeach patient on admission.Results: Estimating the receiver operating characteristic area under the curve showed that RDWhas very gooddiscriminative power for mortality (area under the curve = 0.817; 95% confidence interval, 0.689-0.946).With a cutoff value of 14.8 for RDW, mortality could be correctly predicted in approximately 77% of cases.Conclusions: Red cell distribution width on admission is a predictor of mortality in patients with AP.

© 2013 Elsevier Inc. All rights reserved.

1. Introduction

Acute pancreatitis (AP) is one of the most common surgicalemergencies. Acute pancreatitis ranges from a mild, self-limiteddisease to severe, and sometimes fatal, disorder. Several studiesindicated that the mortality rate of patients with AP is currentlyapproximately 3.8% to 7%; in severe AP, it varies from 7% to 42% [1].Early identification of patients at greater risk of mortality maydetermine a more rational use of diagnostic studies and prompt earlyinstitution of interventional or medical treatment, leading todecreased mortality rates [2]. In previous studies, several biologicalmarkers and clinical events had been used to predict the mortality ofAP [1].

Red cell distribution width (RDW) is a quantitative measure ofvariability in the size of circulating erythrocytes, with higher valuesreflecting greater heterogeneity in cell sizes (ie, anisocytosis). R celldistribution width is an easy, inexpensive, routinely reportedparameter as a part of the complete blood count test used in theassessment of the patient's disorder. Recent studies exhibits that RDWis a remarkable prognostic marker to determine the risk of mortalityin a wide range of clinical manifestations such as community-dwelling older adults with or without age-associated diseases,critically ill patients, intensive care unit patients, and those withcardiovascular diseases (ie, heart failure, coronory diseases) and evenacute dyspnea and community-acquired pneumonia [3–9].

: +90 312 310 34 60.

l rights reserved.

The aim of this study was to examine whether RDW on admissioncould be a predictor of mortality in patients with AP.

2. Materials and methods

This was an observational cohort study. We analyzed 102 cases ofAP admitted between January 2010 and June 2012. The pancreatitisdiagnosis was based on typical clinical presentation, plasma amylaselevel exceeding 3 times the upper normal threshold (120 IU/L), and/orverified pancreatitis by abdominal computed tomographic scanningor ultrasonography.

Demographic data, etiology of pancreatitis, organ failure, meta-bolic disorder, hospitalization time, and the following laboratorymeasures were obtained from each patient record on admission:serum amylase, C-reactive protein, RDW, white blood cell (WBC)count, platelets, hematocrit, calcium, blood glucose, creatinine, bloodurea nitrogen (BUN), arterial and venous blood gas analysis (partialpressure of oxygen, pH, and base excess), total bilirubin, alkalinephosphatase, lactate dehydrogenase, γ-glutamyl transferase, alanineaminotransferase, and aspartate aminotransferase).

The primary end point was hospital mortality.

2.1. Statistical analyses

Data were tested for normality and were found to be nonnormallydistributed. Accordingly, all data are presented as median value(interquartile range), with nonparametric analyses being used toassess differences. Univariate analysis was performed using theMann-Whitney U test and χ2 test when appropriate. Variables on

1,0 0,80,60,40,20,0

1 - Specificity

1,0

0,8

0,6

0,4

0,2

0,0

ROC Curve

Fig. Receiver operating characteristics of the RDW.

688 K. Şenol et al. / American Journal of Emergency Medicine 31 (2013) 687–689

admission found to be significant on univariate analysis were enteredinto multivariate logistic regression analysis.

We measured the prognostic performance of the variables usingreceiver operating characteristic curves and calculated sensitivity,specificity, positive predictive value (PPV) and negative predictivevalue (NPV) using different cutoff values.

All statistical procedures were performedwith SPSS 15.0 (SPSS Inc,Chicago, Illinois). P b .05 was considered significant.

3. Results

There were 59 female and 43 male patients, with a median age of56. 5 (17-89) years. The etiology of AP included 27 alcohol inducedand 75 gallstone induced. Twenty-five of the patients developedcomplications: 15 pseudocysts, 4 pancreatic abscesses, and 6necrotizing pancreatitis. Median hospital stay time was 6 (1-88) days.

Thirteen patients died during the hospital stay. Demographic,clinical, and laboratory characteristics of survivors and nonsurvivorsare summarized in Table 1.

We were able to identify several risk factors on admission thatpredict mortality in univariate analysis (age, WBC count, plateletcount, calcium level, BUN level, RDW, and albumin level). Variables onadmission found to be significant in univariate analysis were enteredinto multivariate analysis. In multivariate analysis, RDWwas the onlyadmission variable that predicted mortality (Table 1).

Receiver operating characteristic curves of RDW levels were usedto identify nonsurvivors on a statistically significant level (area underthe curve of 0.817; 95% confidence interval [CI], 0.689-0.946) (Fig.).We further analyzed the PPV and NPV of high RDW. With a cutoffvalue of 14.8 for RDW, mortality could be predicted in approximately77% of cases (Table 2).

4. Discussion

We have demonstrated that elevated RDW at admission is anindependent risk factor for mortality in AP. This is the first study toshow this significant association between RDW and mortality inpatients with AP.

A great deal of research has focused on development ofapproaches to early risk stratification in AP. Various approacheshave included clinical scoring systems such as the Ranson criteria,modified Glasgow, and Acute Physiology and Chronic HealthEvaluation II score. Additional approaches have included use ofinflammatory markers such as C-reactive protein or interleukin-6and interleukin-8. Prior studies have also evaluated the potentialrole of specific routine laboratory tests in predicting outcome in AP.

Table 1Summary of the demographic and clinical characteristics of the patients

Nonsurvivors (n = 13)

Age 52(41-72)Sex (M/F) 4/9BUN (mg/dL) 20.56 (12.29-48.6)RDW (%) 15.6 (14-21)Platelet count (thousand per μL) 193 (173-212)Amylase (U/L) 1223 (820-2860)WBC count (thousand per μL) 14.4 (12.9-16.)Albumin (g/L) 29 (22-31)Total bilirubin (mg/dL) 2 (1.5-2.7)Lactate dehydrogenase (U/L) 274 (185-331)Alanine aminotransferase (U/L) 77 (29-108)Aspartate aminotransferase (U/L) 121 (38-186)Calcium (mg/dL) 7.84 (7.39-8.49)Serum glucose (mg/dL) 149 (128-200)Hospital stay 3 (2-19)

Data are presented as median value (interquartile range).Bold indicates statistical significance.

However, none of these approaches has gained widespreadacceptance in clinical practice [10].

Mortality in AP is associated with the multiorgan failure and septiccomplications. In recent years, multiorgan failure has been acknowl-edged as a major determinant of mortality that correlates with themagnitude of inflammatory response.Mortality has 2 peaks, ie, “early”during the first week, when the systemic inflammatory responsesyndrome and multiple organ dysfunction syndrome develop, and“late” after 1 to 3 weeks, when mortality often is caused by multipleorgan dysfunction syndrome together with infections and sepsis [11].

For many medical professionals who thought that everything wasalready known about the complete blood count panel and specificallyabout RDW, studies of RDW as a prognosticator of mortality weresurprising. Further studies evaluated RDW for mortality or adverseevents in various cardiac diseases; cancer; stroke; and peripheralarterial, renal, infectious, and pulmonary disease and in critically illpatients as well as general outpatients and general populationsamples [7–9,12].

Previous studies adjusted the association of RDW with adverseevents for a multiplicity of covariables including demographics,anthropometrics, diagnoses and comorbidities, laboratory variables,

Survivors (n = 89) Univariate P Multivariate P

77(73-82) P = .001 NS39/50 NS12.66 (10.28-17.76) P = .001 NS13.3 (12.5-14.3) P = .000 P = .001236 (197-298) P = .037 NS1114 (614-1663) NS11.4 (8.2-14.3) P = .018 NS35 (30-38) P = .001 NS2 (.90-3.60) NS301 (195-481) NS118 (30-291) NS122 (41-253) NS8.75 (8.32-9.13) P = .001 NS138 (112-172) NS6 (3-9) NS

Table 2Sensitivity, specificity, +LR and −LR, PPV, and NPV of RDW

RDW Sensitivity(95% CI)

Specificity(95% CI)

+LR −LR PPV(95% CI)

NPV(95% CI)

N14.8 47.6(38.3-57.9)

96.3 (91.8-99) 12.8 0.54 76.9%(48.9-93.5)

87.6%(83.6-90.1)

+LR indicates positive likelihood ratio; −LR, negative likelihood ratio.

689K. Şenol et al. / American Journal of Emergency Medicine 31 (2013) 687–689

behavioral factors, nutritional parameters, and medical treatments.Those studies intended to discover whether RDW was simply amarker for other known risk factors, which could help explain why itpredicts risk. The evidence demonstrated, however, that statisticaladjustment for each of those factors failed to eliminate the predictiveability of RDW [12].

The pathophysiologic mechanisms for the association betweenhigher RDW and mortality in AP are unclear. Inflammation may helpexplain the underlying association of RDW with mortality. Inflam-mation influences bone marrow function and iron metabolism, andinflammatory cytokines suppress erythrocyte maturation, accentuat-ed with sepsis, allowing newer, larger reticulocytes to enter thecirculation, which is associated with RDW increase [5]. High oxidativestress can also lead to elevated RDW by reducing red blood cell (RBC)survival and increasing release of large premature RBCs into theperipheral circulation. In addition, inflammation alters RBC mem-brane glycoproteins and ion channels, contributing to the change ofRBC morphology [13,14]

There are several limitations to this study. Our study is mainlylimited by the retrospective nature and the small number of patients.Second, despite a single measurement of RDW level that could havereflected acute changes in RDW induced by blood loss or hemolysis,we did not evaluate fluctuations in RDW levels and thus could notaccount for possible variation over time. Third, RDW samples werecollected from a single center. The value of RDW in the prediction ofclinical outcomes may have been slightly different if the populationwas different. Further prospective studies with larger patientpopulations involving multiple centers are necessary to moreaccurately assess high RDW as a predictor of mortality.

In conclusion, increased RDW at admission was an independentpredictor of mortality in patients with AP. Furthermore, these datasuggest that clinicians can use increased RDW as a prognostic markerfor AP patients and should pay additional attention to patients withincreased RDW.

References

[1] Wang X, Cui Z, Zhang J, Li H, Zhang D, Miao B, et al. Early predictive factors of inhospital mortality in patients with severe acute pancreatitis. Pancreas 2010;39:114–5.

[2] Yoldaş O, Koç M, Karaköse N, Kiliç M, Tez M. Prediction of clinical outcomes usingartificial neural networks for patients with acute biliary pancreatitis. Pancreas2008;36:90–2.

[3] Patel KV, Semba RD, Ferrucci L, et al. Red cell distribution width and mortality inolder adults: a meta-analysis. J Gerontol A Biol Sci Med Sci 2010;65:258–65.

[4] Patel KV, Ferrucci L, Ershler WB, Longo DL, Guralnik JM. Red blood cell distributionwidth and the risk of death in middle-aged and older adults. Arch Intern Med2009;169:515–23.

[5] Ku NS, Kim HW, Oh HJ, et al. Red blood cell distribution width is an independentpredictor of mortality in patients with gram-negative bacteremia. Shock 2012;38:123–7.

[6] Hunziker S, Celi LA, Lee J, Howell MD. Red cell distribution width improves thesimplified acute physiology score for risk prediction in unselected critically illpatients. Crit Care 2012;16:R89.

[7] Makhoul BF, Khourieh A, Kaplan M, Bahouth F, Aronson D, Azzam ZS. Relationbetween changes in red cell distribution width and clinical outcomes in acutedecompensated heart failure. Int J Cardiol 2012 [Epub ahead of print] PubMedPMID: 22560496.

[8] Hong N, Oh J, Kang SM, Kim SY, Won H, Youn JC, et al. Red blood cell distributionwidth predicts early mortality in patients with acute dyspnea. Clin Chim Acta2012;413:992–7.

[9] Braun E, Domany E, Kenig Y, Mazor Y, Makhoul BF, Azzam ZS. Elevated red celldistribution width predicts poor outcome in young patients with communityacquired pneumonia. Crit Care 2011;15:R194.

[10] Wu BU, Bakker OJ, Papachristou GI, et al. Blood urea nitrogen in the earlyassessment of acute pancreatitis: an international validation study. Arch InternMed 2011;171:669–76.

[11] Carnovale A, Rabitti PG, Manes G, Esposito P, Pacelli L, Uomo G. Mortality in acutepancreatitis: is it an early or a late event? JOP 2005;6:438–44.

[12] Horne BD. A changing focus on the red cell distribution width: why does it predictmortality and other adverse medical outcomes? Cardiology 2012;122:213–5.

[13] Ghaffari S. Oxidative stress in the regulation of normal and neoplastichematopoiesis. Antioxid Redox Signal 2008;10:1923–40.

[14] Song CS, Park DI, YoonMY, Seok HS, Park JH, Kim HJ, et al. Association between redcell distribution width and disease activity in patients with inflammatory boweldisease. Dig Dis Sci 2012;57:1033–8.