Comparing Automated Vs Manual Leukocyte Differential Counts for Quantifying the ‘Left Shift’ in the Blood of Neonates

Home , Left shift (medicine)

ORIGINAL ARTICLE Comparing automated vs manual leukocyte differential counts for quantifying the ‘left shift’ in the blood of neonates

BC MacQueen1, RD Christensen1,2,3, BA Yoder1,2, E Henry2,4, VL Baer1,2, ST Bennett5 and HM Yaish3

OBJECTIVE: The neutrophil ‘left shift’ can be measured via the immature to total (I/T) neutrophil ratio or the absolute bands per μl using a manual differential count. It can also be measured from an automated differential count by the immature granulocyte percentage (IG%) or the absolute IG per μl. In neonates, it is unknown if the manual or automated differential count is superior. STUDY DESIGN: We directly compared complete blood counts (CBCs) with manual and automated differential counts from infants o90 days old, and documented whether or not each neonate was infected. We developed reference intervals for I/T ratio, bands per μl, IG% and IG per μl using values from non-infected neonates. RESULTS: The database had 10 714 CBCs. The upper reference interval for I/T ratio was 0.29 in the ﬁrst 48 h and 0.31 thereafter; bands per μl was 3710 μl − 1 in the ﬁrst 48 h and 1785 μl − 1 thereafter. IG% was 6.2% then 4.2%; IG per μl was 1460 μl − 1 then 613 μl − 1. Statistical performances of the four methods were equivalent for identifying infection. CONCLUSIONS: We developed reference intervals for four methods of quantifying a neonate’s ‘left shift’. The information from automated differentials is not inferior to that from manual differentials in identifying infections, but automated differentials have the advantages of a larger sample size, being less expensive, and faster performance times. Journal of Perinatology (2016) 36, 843–848; doi:10.1038/jp.2016.92; published online 9 June 2016

INTRODUCTION on the basis of a much larger sample of leukocytes enumerated, Neutrophil ‘left shift’ is an expression used to indicate an abnormal plus removal of human error in discriminating among cell types, increase in immature neutrophils in the circulation.1–3 One because cell-type assignment is determined automatically by 11 method of quantifying the ‘left shift’, particularly popular in preset gating techniques. neonatal hematology, is the immature to total (I/T) neutrophil In many clinical laboratories, automated differential counts ratio.1–6 This requires a manual differential cell count where a have replaced manual counts for adult patients, but it is not clear hematology technologist microscopically examines 100 leuko- whether this is appropriate for complete blood counts (CBCs) of 10,12 cytes on a stained blood film, enumerating each cell accord- newborn infants. A barrier to adopting automated leukocyte ing to morphological characteristics. The I/T ratio is typically differentials in neonatology is the lack of reference intervals for 13 calculated as the percent immature neutrophils divided by the the IG% and IG per μl in neonatal populations. Also, the percent segmented neutrophils+band neutrophils+metamyelo- usefulness of the IG% and the IG per μl as biomarkers for infection cytes. A second common method of quantifying the ‘left shift’ is have not been compared directly with the I/T ratio and the the absolute band count, which also requires a manual differential. absolute band count in infected vs non-infected neonates. The percent of leukocytes identified as bands is multiplied by We designed the present study for three related purposes: (1) to the leukocyte count and the value expressed as bands per μlof establish reference intervals for IG% and IG per μl from automated blood.7 differential counts using a large sample size of non-infected The automated leukocyte differential cell count is a relatively neonates from birth to 90 days old, (2) to determine paired new innovation in clinical laboratory hematology, where large correlations between the IG% from automated differentials and numbers of leukocytes are categorized by flow cytometric the I/T ratio from manual differentials, and also between the techniques according to their size and cytoplasmic and nuclear absolute IG per μl (automated) and the absolute band count characteristics.4,8,9 In an automated differential cell count on some (manual), and (3) to compare the statistical performance for models of Sysmex hematology analyzers, the ‘left shift’ is identifying infection by these four methods of quantifying the ‘left quantified by ‘immature granulocytes (IG)’. The IG% and the IG shift’, using CBCs from large numbers of infected vs non-infected per μl are somewhat analogous to, but technically distinct from, neonates. the I/T ratio and the absolute band count. Automated differentials We speculated that with this new information we could make a have the advantage that a blood smear is not needed, and data-based decision whether to change the CBC method for technician time is not required to perform a microscopic analysis, neonates in the Intermountain Healthcare and University of Utah thereby conserving materials and reducing technician time.9,10 healthcare systems from manual to automated differentials, thus Moreover, enhanced performance of the automated differential potentially improving accuracy and lowering operating costs of over manual methods has been reported from adult populations the test.

1Division of Neonatology, Department of Pediatrics, University of Utah School of Medicine, Salt Lake City, UT, USA; 2Women and Newborn’s Clinical Program, Intermountain Healthcare, Salt Lake City, UT, USA; 3Division of Hematology/Oncology, Department of Pediatrics, University of Utah School of Medicine, Salt Lake City, UT, USA; 4Institute for Healthcare Delivery Research, Salt Lake City, UT, USA and 5Intermountain Medical Center, Department of Pathology, Murray, UT, USA. Correspondence: Dr BC MacQueen, Division of Neonatology, Department of Pediatrics, University of Utah School of Medicine, Salt Lake City, UT 84108, USA. E-mail: [email protected] Received 9 February 2016; revised 1 April 2016; accepted 26 April 2016; published online 9 June 2016 Leukocyte ‘left shift’ in neonates BC MacQueen et al 844 METHODS records were transferred to an electronic data file accessed only by The study protocol was approved by the Intermountain Healthcare Intermountain Healthcare research data analysts (directed by EH). Institutional Review Board. A waiver from individual parental consent All blood cell counts were determined using Sysmex hematology μ was granted because this was a deidentified data-only study with analyzers. IG% and IG per l were automatically determined by using fl 1,14 appropriate privacy protection. Intermountain Healthcare is a not-for- polymethine dye, uorescence and light scattering analysis. All blood profit healthcare system operating 18 hospitals with labor and delivery tests were performed in accordance with Intermountain Healthcare ’ units in Utah and Idaho. Laboratory Services standard operating procedures and manufacturer s Hematology analyzers used in larger Intermountain Healthcare clinical instructions. The Sysmex quality control procedures were performed daily laboratories generate IG% and IG per μl on every CBC with white blood cell as recommended by the manufacturer. differential order, but those values are reported to the clinical record only Reference intervals (5th and 95th percentile lower and upper limits) were developed for IG% and IG per μl after excluding from the database all when the order is for a CBC with automated differential, and the IG% and values from neonates likely to have an infection. This was accomplished by IG per μl results are not flagged by the analyzer as potentially invalid. identifying neonates with a positive culture of a usually sterile body fluid A ‘flag’ is an indication from the instrument that a reported result is (specifically, blood, urine or spinal fluid) and also those with a negative abnormal. This means that either a result is outside the established normal culture but a diagnosis of ‘clinical sepsis’ defined by antibiotic adminis- ranges programmed into the instrument or the instrument detects a tration for 45 days because of the diagnosis of pneumonia or clinical technical problem with the sample. For this study each CBC from neonates μ sepsis charted by the attending physician. up to 90 days old, including the IG% and IG per l, were periodically The statistical performance of each of the four means of quantifying the manually downloaded from Sysmex (Sysmex, Kobe, Japan) hematology left shift was assessed by sensitivity, specificity, positive predictive value fi analyzers to delimited research-only text les. I/T ratio and absolute band and negative predictive value (Statit (Corvallis, OR). Receiver operating count results were obtained from the Intermountain Healthcare Enterprise characteristic curve analyses were performed using IBM SPSS Statistics, Data Warehouse. The I/T ratio was calculated from the manual leukocyte Version 23.0, Armonk, NY, USA. differential as the percent immature neutrophils divided by the percent segmented neutrophils+band neutrophils+metamyelocytes (Figure 1). The absolute band count per μl was calculated from the percent band RESULTS neutrophils on the manual differential count multiplied by the white blood Study data sets cell count. The IG% and IG per μl were reported directly from the automated leukocyte differential. The IG% consists of the percent The original data set had 11 986 values. Of these, 1272 (10.6%) had promyelocytes+myelocytes+metamyelocytes (Figure 1). The IG per μlis a ‘flag’ and were excluded from the study leaving 10 714 CBCs the IG% multiplied by the white blood cell count. with simultaneous automated and manual differential cell counts Records were excluded from the study set if IG%, IG per μl, I/T ratio or for the study data set. These CBCs were obtained from 16 absolute band count results were unavailable or if the IG% or IG per μl analyzers in 10 Intermountain Healthcare hospital laboratories results were flagged by the analyzer as requiring confirmation. Study between 29 May 2014 and 22 September 2015 (see Table 1 for additional information). The 10 714 CBCs came from 5281 individual neonates with an average of 2 CBCs per neonate. Over 50% (2662/5281) of the samples came from individual neonates, and 90% (4737/5281) of the samples came from neonates contributing 1, 2 or 3 CBC samples. The gestational age of the neonates whose CBCs were included was 37.6 ± 2.4 (mean ± s.d.) weeks. The great majority of the samples (91%) were from neonates ⩾ 34 weeks gestation. The data were divided into two additional sets on the basis of presence or absence of infection. Reference intervals (5th to 95th percentile lower and upper limits) were created from CBCs of infants who did not have evidence of Figure 1. Neutrophil maturation. Myeloblasts, promyelocytes and infection or Trisomy 21. myelocytes are capable of cell division and together are referred to as mitotic neutrophils. Metamyelocytes, bands and segmented Correlations of ‘left shift’ values from manual vs automated neutrophils are not capable of cell division and together are referred leukocyte differential counts to as post-mitotic neutrophils. In an automated differential cell count, the IG% is the percent of circulating neutrophils made up of The I/T ratio and IG% from all 10 714 CBCs of infants between promyelocytes, myelocytes and metamyelocytes. In a manual birth and 90 days old are shown in Figure 2. The I/T ratio and IG% differential cell count, the I/T ratio is the percent of post-mitotic had a weak positive linear relationship (r = 0.18; Po0.0001). The neutrophils made up of metamyelocytes and band neutrophils. absolute bands per μl and the IG per μl from all 10 714 CBCs had a

Table 1. Shown here are the Intermountain Healthcare clinical laboratories where the 10 714 CBCs were run. Also shown are the cities where each laboratory is located, their altitudes relative to sea level, the Sysmex analyzer model used and number of values contributed to the data set from that site

Clinical laboratory City Altitude (ft/m above sea level) Sysmex model Number (CBC values obtained from that site)

Alta View Hospital Sandy 4797 ft, 1462 m XT-4000i 98 American Fork Hospital American Fork 4610 ft, 1405 m XE-5000 889 Dixie Regional Medical Center St. George 2661 ft, 881 m XE-5000 97 Intermountain Medical Center Murray 4500 ft, 1372 m XE-5000 3829 LDS Hospital Salt Lake City 4587 ft, 1398 m XE-5000 908 Logan Regional Logan 4573 ﬂ, 1394 m XE-5000 488 McKay-Dee Hospital Center Ogden 4649 ft, 1417 m XE-5000 1549 Primary Children’s Hospital Salt Lake City 5003 ft 1525 m XE-5000 1338 Riverton Hospital South Riverton 4439 ft, 1353 m XT-4000i 14 Utah Valley Regional Medical Center Provo 4590 ft, 1399 m XE-5000 1507

Journal of Perinatology (2016), 843 – 848 © 2016 Nature America, Inc., part of Springer Nature. Leukocyte ‘left shift’ in neonates BC MacQueen et al 845 moderate positive linear relationship, as shown in Figure 2 thereafter.1,2,13 Therefore, the reference intervals we generated (r = 0.45, Po0.0001). were divided into two periods of time: first 48 h after birth, and 48 h through 90 days. As shown in Table 2, the upper reference limit (95th percentile value) for I/T ratio during the first 48 h was Neonatal reference intervals for automated differential cell counts 0.29; thereafter it was essentially the same (0.31). The upper Reference intervals were generated using the CBCs of neo- reference limit for IG% during the first 48 h after birth was 6.2% nates with no evidence of infection. Blood neutrophil concentra- and thereafter was 4.2%. The upper reference limit for absolute tions are typically higher during the first 48 h after birth than band count during the first 48 h was 3710 per μl and thereafter was 1785 per μl. The upper reference limit for absolute IG count during the first 48 h was 1460 per μl and thereafter was 613 per μl. Lower reference limits were essentially zero in all cases. Reference interval charts for IG% and IG per μl for the first week after birth are shown in Figure 3.

Discrepancies between the IG% and the I/T ratio Six-hundred and ﬁve CBCs were available from neonates with an elevated IG% but a normal I/T ratio. As shown in Table 3, 44% of these were diagnosed with and treated for an infection, despite the normal I/T ratio. Conversely, 830 CBCs had a normal IG% but an elevated I/T ratio, and 49% of these were diagnosed with and treated for infection. Those with elevations in both IG% and I/T (n=129) had a higher likelihood (66%) of receiving treatment for infection.

Left shift measurements in neonates with Trisomy 21 We did not have an a priori aim related to the ‘left shift’ in neonates with Trisomy 21. However, on examining the records of neonates with an elevated IG% (to obtain the data for Table 3), it appeared to us that the diagnosis of Trisomy 21 was overrepresented in this group. On the basis we assembled the data shown in Table 4, which illustrates that indeed non-infected neonates with Trisomy 21 had a higher ‘left shift’, measured by of the four parameters, than did non-infected neonates without Trisomy 21. Moreover the group of neonates with Trisomy 21 with infections had higher ‘left shift’ measurements than did the group of infected neonates who did not have Trisomy 21.

Statistical performance of left shift measurements for identifying neonatal infection These four methods of quantifying the ‘left shift’ had similar statistical performance for suggesting the diagnosis of infection (Table 5). For each of the four, the sensitivity was weak (12 to 15%), whereas the speciﬁcity was strong (90 to 95%), and the positive predictive value and negative predictive value were ’ a typically in the low to mid 60 s%. Combinations of the four Figure 2. ( ) I/T ratio vs IG%. Relationship between I/T ratios from tended to diminish the sensitivity but increased the positive manual differential cell counts and IG% from automated differential counts. Paired results are shown as individual dots for n=10 714 predictive value. CBCs (r = 0.18). (b) Absolute band count vs IG per μl. Relationship Receiver operating characteristic curve analysis for each between absolute and counts from manual differential cell counts parameter was assessed for the three outcomes shown in and IG from automated differential counts. Paired results are shown Table 3 (culture-positive infection, culture-negative clinical infec- as individual dots for n=10 714 CBCs (r = 0.45). tion and these two outcomes combined). Area under the curve

Table 2. Reference intervals for IG%, I/T ratio, IG per μl and bands per μl in non-infected neonates (n=6462 CBCs) during the ﬁrst 90 days after birth

First 48 h after birth (n = 4808) Beyond 48 h (n = 1654)

Measurement Median ± s.d. 5th percentile 95th percentile Median ± s.d. 5th percentile 95th percentile

IG% 1.99 ± 1.70 0.50 6.20 1.62 ± 0.70 0.20 4.20 I/T ratio 0.09 ± 0.08 0 0.29 0.12 ± 0.05 0 0.31 IG (cells per μl) 539 ± 270 50 1460 308 ± 70 10 613 Bands (cells per μl) 1303 ± 680 0 3710 702 ± 160 0 1785 Abbreviations: IG, immature granulocyte; I/T, immature to total neutrophil ratio. The upper reference limits are shown in shaded columns for two periods: the ﬁrst 48 h after birth (ﬁrst shaded column) and the beyond 48 h (last shaded column).

© 2016 Nature America, Inc., part of Springer Nature. Journal of Perinatology (2016), 843 – 848 Leukocyte ‘left shift’ in neonates BC MacQueen et al 846 values never exceeded 0.65 for any individual parameter or in the blood. A traditional method of quantifying the ‘left shift’, combination of parameters tested. particularly popular in Neonatology, has been the I/T neutrophil ratio.1–6,16 Criticisms of the accuracy of the I/T ratio include the difficulty of even well-trained technologists to reliably differentiate DISCUSSION between band and segmented forms, which is an essential profess The expression ‘leukocyte left shift’ was derived from the practice for obtaining an accurate I/T ratio.3,5 No functional differences of tallying leukocytes during a manual differential cell count, with have ever been shown between band vs segmented neutrophils, the more mature neutrophils tallied to the extreme right-side and morphological differences can be minor and arbitrary.7,13 of the tally device and the less mature forms tallied progressively A related criticism of the I/T ratio is the relatively small sample of to the left.2,15 Thus a ‘left shift’ in the manual differential cell count cells enumerated by the technologist performing a manual indicates an abnormally high proportion of immature neutrophils differential cell count.2,3,5 Sample size errors are reduced when more leukocytes are counted in a manual differential, but the usual procedure is to count 100 leukocytes.17,18 However, in any given set of 100 leukocytes, typically fewer than 60 to 70%, and sometimes less than half, are neutrophils. These two problems limit the validity of the I/T ratio. Automated leukocyte differential cell counts have advantages over manual counts, because human discrimination between cell forms is not required and multi-log-higher numbers of cells are enumerated.19–21 However, reference intervals for automated differential counts have not previously been reported for neonates,13 and the statistical performance of automated vs manual differentials for identifying infection has not previously been compared. The earliest morphologically recognizable neutrophils are the myeloblast, promyelocyte and myelocyte. These cells mature into metamyelocytes then bands and finally into segmented neutrophils (Figure 1). Myeloblasts, promyelocytes and myelocytes are capable of cell division, as evidenced by tritiated thymidine incorporation into their nuclear DNA and verified by direct observation in culture.22,23 These three early stages of neutrophils are collectively referred to as the ‘neutrophil mitotic compartment’, or the ‘neutrophil proliferative pool’.23 During infections, the neutrophil mitotic compartment typically adds one or two additional cell divisions, thereby expanding the proliferative pool size.23 Adding one more division doubles the number of mature neutrophils produced. During infection, the number of promyelocytes and myelocytes in the bone marrow typically increases because of the added cell divisions.24 A small fraction of these neutrophil precursors escape the marrow and are found in the blood, and can be measured by the IG%. A separate mechanism for delivering more neutrophils to infected tissues is premature release of immature post-mitotic neutrophils (metamyelocytes and bands) from the marrow into the circulation. This leads to an increase in the I/T ratio.2,16,25–28 In the present study, we found that the I/T ratio and the IG% had a weak positive linear correlation. The weak correlation Figure 3. Reference intervals. 5th, median and 95th percent values can be understood by recognizing that these two measurements are displayed. (a) Reference intervals for IG% over the first week following birth. (b) Reference intervals for IG per μl over the first assess distinct developmental stages of neutrophils. The I/T week following birth. ratio focuses on mature neutrophils, principally evaluating the

Table 3. Explanations for 1564 CBCs where either the IG% or the IT ratio was elevated (495% upper reference limit) in a discrepant way vs both values were elevated

Group number Group description Culture positive Culture negative or Culture positive or Likely a clerical error in infection clinical infection clinical infection manual differentiala

1 Elevated IG% but normal IT ratio (n=605) 42 (6.9%) 224 (37.0%) 266 (44.0%) 0 2 Elevated I/T ratio but normal IG% (n=830) 73 (8.8%) 334 (40.2%) 407 (49.0%) 3 3 Elevated IG% and elevated I/T ratio (n=129) 19 (14.7%) 66 (51.2%) 85 (65.9%) 0 P-value group 1 vs 2 0.202 0.217 0.057 P-value group 1 vs 3 0.004 0.003 0.000 P-value group 2 vs 3 0.003 0.019 0.000 Abbreviations: IG, immature granulocyte; I/T, immature to total neutrophil ratio. aA repeat manual differential within 3 to 6 h was reversed from the ﬁrst (bands were likely initially reported as segs, and segs were likely reported as bands) and the clinician noted in the record that the differential count was likely erroneous.

Table 4. ‘Left shift’ measurements of neonates with Trisomy 21

No infection Infectiona

IG% I/T IG per μl Band per μl IG% I/T IG per μl Band per μl

Trisomy 21 5.12 ± 3.56 0.16 ± 0.14 865 ± 970 1489 ± 1597 8.48 ± 8.33 0.25 ± 0.18 1573 ± 2517 2270 ± 2762 No Trisomy 21 2.03 ± 1.92 0.09 ± 0.01 378 ± 498 926 ± 1213 2.33 ± 2.85 0.15 ± 0.16 443 ± 996 1214 ± 1685 P-value 0.000 0.001 0.000 0.008 0.000 0.000 0.002 0.008 Abbreviations: IG, immature granulocyte; I/T, immature to total neutrophil ratio. aCulture-positive infection or clinical infection (intravenous antibiotics ⩾ 5 days duration). In the ﬁrst row, 61 CBCs from 27 non-infected neonates with Trisomy 21 are shown next to 53 CBCs from 15 infected aneonates with Trisomy 21. In the second row, 6402 CBCs from 3964 non-infected neonates who did not have Trisomy 21 are shown next to 4198 CBCs from 1275 infected aneonates who did not have Trisomy 21. (mean ± s.d.). Vertical comparison between neonates with vs without Trisomy 21 are shown in the third row.

Table 5. Test performance

Measurement Sensitivity (%) Speciﬁcity (%) PPV (%) NPV (%)

Elevated I/T ratio 15.1 94.6 64.7 62.9 Elevated IG% 13.0 94.8 62.4 62.4 Elevated band count 13.1 95.0 63.3 62.4 Elevated IG count 12.4 95.0 61.8 62.2 Elevated I/T and IG% 2.4 89.9 74.8 64.8 Elevated band count and Elevated IG count 4.8 91.2 72.3 63.6 Elevation in all four ‘left shift’ measures 1.6 87.1 78.7 65.5 Abbreviations: IG, immature granulocyte; I/T, immature to total neutrophil ratio; NPV, negative predictive value; PPV, positive predictive value. aCulture- positive infection or clinical infection (intravenous antibiotics X5 days duration). Sensitivity, speciﬁcity, PPV and NPV for identifying infectiona, comparing different measurements of the leukocyte ‘left shift’.

proportion of bands to segmented neutrophils, whereas the IG% and when the CBC was immediately repeated the I/T ratios were focuses on yet less mature neutrophils, quantifying the proportion normal (o0.2). Such errors would not occur with automated of granulocytes composed of promyelocytes, myelocytes and differential counts. Perhaps in cases where the reason for metamyelocytes. obtaining the CBC is as part of a ‘rule out sepsis’ evaluation, a On the basis of what is known about neutrophil kinetics during manual differential count to obtain an I/T ratio, along with an IG% infection,10,15,24 it was not surprising to us to find that some and IG per μl from the automated differential, could give a higher neonates had an elevated IG% and yet had a normal I/T ratio. In positive predictive value and higher negative predictive value. fact, 6.9% of 605 neonates with an elevated IG% and a normal I/T Neonates with Trisomy 21 have altered hematopoiesis leading ratio had a positive culture at the time the CBC was performed, to a high rate of abnormalities in all blood cell lineages. Most and another 37% were treated for a culture-negative clinical infants with Trisomy 21 have neutrophilia in the first week of life.28 infection. In this latter group the elevated IG% could not have Our study further revealed that non-infected infants with Trisomy been a factor in the clinician’s decision to treat with antibiotics for 21 had higher I/T ratios and IG% than non-infected infants without 45 days, because the IG% values were not released to the Trisomy 21. Clinicians should be aware of this phenomenon when clinicians in this study but were shunted to a research-only file. interpreting CBC results from infants with Trisomy 21. Thus, something else about these neonates with a normal I/T ratio We recognize that our study has several limitations. First, prompted the clinicians to diagnose an infection. although 10 714 CBCs were electronically investigated and results Our purpose in performing this study was not to identify the analyzed, only 1564 of these had the charts reviewed individually. best means of rapidly identifying neonatal infection; in a This could result in less accuracy in identifying those individuals study with that aim we might have tested combinations of CRP, with definite infection. Second, when defining infection in the procalcitonin, leukocyte count and means of quantifying the ‘left infants, we recorded a positive culture of a usually sterile body shift’.6 Rather, our focus was on establishing reference intervals for fluid or a diagnosis of ‘clinical sepsis’ which is subjective. We did and comparing the ‘left shift’ measures in automated vs manual not attempt to identify viral processes such as bronchiolitis which differential counts. We found that the statistical performance of might affect the neutrophil differential. Also, we were unable to the I/T ratio and IG%, in identifying neonates with infection, was determine the duration of abnormal ‘left shift’ measurements, and essentially identical. we are uncertain when each abnormal value occurred; early, As previously reported, the I/T ratio is not a sensitive method of middle or late in the course of the infection. Third, some of the identifying infection, but it is fairly specific for infection. The IG per cases we labeled as infected might have had a positive culture μl and band counts performed similar to the I/T ratio and IG%. that was actually a contaminant. Perhaps not all of these were These findings lead us to conclude that for most purposes recognized as a contaminant and removed from the infected data an automated differential should serve as well as a manual set. Fourth, we were not able to determine with certainty whether differential, similar to conclusions drawn from CBC obtained from each CBC, in a neonate with infection, was early, mid or late during adults.12,20 Moreover, we found three manual differential counts the antibiotic course. This undoubtedly lowers the sensitivity of that had what we surmise were transposition errors (bands placed the left shift parameters to identify infection. Finally, during our in the seg column, and vice versa). Reviewing these cases, they review of the electronic records, we recognized that Trisomy had extraordinarily high I/T ratios (40.8) and were well appearing, 21 was overrepresented with high I/T and IG% without evidence

© 2016 Nature America, Inc., part of Springer Nature. Journal of Perinatology (2016), 843 – 848 Leukocyte ‘left shift’ in neonates BC MacQueen et al 848 of infection. There could be other unidentified groups of infants 6 Newman TB, Draper D, Puopolo KM, Wi S, Escobar GJ. Combining immature and within our data set with inherent abnormalities in their neutrophil total neutrophil counts to predict early onset sepsis in term and late preterm counts, which could skew our reference intervals away from truly newborns: use of the I/T2. Pediatr Infect Dis J 2014; 33:798–802. 22 – normal values. 7 Cornbleet PJ. Clinical utility of the band count. Clin Lab Med 2002; :101 136. The complete blood count with a leukocyte differential is one 8 Arneth BM, Ragaller M, Hommel K, Tiebel O, Menschikowski M, Siegert G. Novel parameters of extended complete blood cell count under fluorescence flow of the most common laboratory tests ordered on neonates and cytometry in patients with sepsis. J Clin Lab Anal 2014; 28: 130–135. young infants. We conclude that as a way to quantify the 9 Becker PH, Fenneteau O, Da Costa L. Performance evaluation of the Sysmex leukocyte ‘left shift’ the IG% and IG per μl from an automated XN-1000 hematology analyzer in assessment of the white blood cell count dif- differential count on the Sysmex hematology analyzer are ferential in pediatric specimens. Int J Lab Hematol 2016; 38:54–63. comparable to the I/T ratio and absolute band count based on a 10 Wong EC. Hematology analyzers: special considerations for pediatric patients. Clin 35 – manual differential count. Automated differential leukocytes using Lab Med 2015; : 165 181. 11 Chabot-Richards DS, George TI. White blood cell counts: reference methodology. other types of cell counters should also be compared with manual Clin Lab Med 2015; 35:11–24. differential counts. On the basis of our present findings, we 12 Kim H, Hur M, Choi SG, Moon HW, Yun YM, Hwang HS et al. Performance eva- conclude that an automated differential count should suffice for luation of Sysmex XN hematology analyzer in umbilical cord blood: a comparison neonatal medicine. For selected neonates where identifying study with Sysmex XE-2100. Clin Chem Lab Med 2014; 52: 1771–1779. infection is the purpose of the CBC, adding a manual differential 13 Wiland EL, Sandhaus LM, Georgievskaya Z, Hoyen CM, O'Riordan MA, Nock ML. count could slightly enhance the test performance. Adult and child automated immature granulocyte norms are inappropriate for evaluating early-onset sepsis in newborns. Acta Paediatr 2014; 103:494–497. 14 Nigro KG, O'Riordan M, Molloy EJ, Walsh MC, Sandhaus LM. Performance of an CONFLICT OF INTEREST automated immature granulocyte count as a predictor of neonatal sepsis. Am J Clin Pathol 2005; 123: 618–624. fl The authors declare no con ict of interest. 15 Boggs DR, Marsh JC, Chervenick PA, Cartwright GE, Wintrobe MM. Neutrophil releasing activity in plasma of normal human subjects injected with endotoxin. Proc Soc Exp Biol Med 1968; 127:689–693. ACKNOWLEDGEMENTS 16 Christensen RD, Shigeoka AO, Hill HR, Rothstein G. Circulating and storage neu- We thank Theodore J. Pysher, Professor of Pathology, University of Utah and Director trophil changes in experimental type II group B streptococcal sepsis. Pediatr Res Of Clinical Pathology Primary Children’s Hospital, and Christopher M. Lehman, 1980; 14: 806–808. Professor of Pathology University of Utah and Medical Director Of Clinical Pathology, 17 Novak RW. The beleaguered band count. Clin Lab Med 1993; 13: 895–903. University of Utah Hospital for helpful discussions and for reviewing the manuscript. 18 Gagon TE, Athens JW, Boggs DR, Cartwright GE. An evaluation of the variance of We also thank Dr. Jolanta Kunicka and Lance Kocherhans of Sysmex America Inc. for leukocyte counts as performed with the hemocytometer, Coulter, and Fisher 36 – helpful discussions. We also thank the following members of the Intermountain intruments. Tech Bull Regist Med Technol 1966; :274 281. Healthcare Clinical Laboratory staff for extracting data from the hematology 19 Bernstein LH, Rucinski J. Measurement of granulocyte maturation may improve 49 – analyzers; Robert Childs, Dereck Daniels, Robert Dallin, Rachael Ethington, Dave the early diagnosis of the septic state. Clin Chem Lab Med 2011; : 2089 2095. 20 Fernandes B, Hamaguchi Y. Automated enumeration of immature granulocytes. Garner, Laurie Jeffs, Jennessa Jessop, Heather Kuusela, Chris Partington, Tim Patten, Am J Clin Pathol 2007; 128: 454–463. Brooke Qi, Lauralyn Ragin, Tara Regginello, Aba Saunders, Alicia Simon, Jennifer 21 Ha SO, Park SH, Park SH, Park JS, Huh JW, Lim CM et al. Fraction of immature Sorensen, Thomas Taylor, Ashley Windsor and Ryan Wilcox. The study was supported granulocytes reflects severity but not mortality in sepsis. Scand J Clin Lab Invest by a research grant (to RDC) from the Intermountain Research and Medical 2015; 75:36–43. Foundation, Murray, UT, USA. 22 Walker RI, Willemze R. Neutrophil kinetics and the regulation of granulopoiesis. Rev Infect Dis 1980; 2:282–292. REFERENCES 23 Skubitz KM. Neutrophilic leukocytes. In: Greer JP, Arber DA, List AF, Glader B, Means RT, Paraskevas F et al. Wintrobe's Clinical Hematology. Lippincott, Williams & 1 Manroe BL, Rosenfeld CR, Weinberg AG, Browne R. The differential leukocyte Wilkins: Philadelphia, 2014, 2003, p137. count in the assessment and outcome of early-onset neonatal group B strepto- 24 Marsh JC, Boggs DR, Cartwright GE, Wintrobe MM. Neutrophil kinetics in acute coccal disease. J Pediatr 1977; 91: 632–637. infection. J Clin Invest 1967; 46: 1943–1953. 2 Christensen RD, Bradley PP, Rothstein G. The leukocyte left shift in clinical and 25 Engle WA, McGuire WA, Schreiner RL, Yu PL. Neutrophil storage pool depletion in experimental neonatal sepsis. J Pediatr 1981; 98:101–105. neonates with sepsis and neutropenia. J Pediatr 1988; 113:747–749. 3 Greenberg DN, Yoder BA. Changes in the differential white blood cell count in 26 Cairo MS. Neutrophil storage pool depletion in neonates with sepsis. J Pediatr screening for group B streptococcal sepsis. Pediatr Infect Dis J 1990; 9:886–889. 1989; 114:1064–1065. 4 Lawrence SM, Eckert J, Makoni M, Pereira HA. Is the use of complete blood counts 27 Henry E, Walker D, Wiedmeier SE, Christensen RD. Hematological abnormalities with manual differentials an antiquated method of determining neutrophil during the first week of life among neonates with Down syndrome: data from a composition in newborns? Ann Clin Lab Sci 2015; 45: 403–413. multihospital healthcare system. Am J Med Genet A 2007; 143A:42–50. 5 Schelonka RL, Yoder BA, Hall RB, Trippett TM, Louder DS, Hickman JR et al. 28 Wheeler JG, Chauvenet AR, Johnson CA, Dillard R, Block SM, Boyle R et al. Neu- Differentiation of segmented and band neutrophils during the early trophil storage pool depletion in septic, neutropenic neonates. Pediatr Infect Dis newborn period. J Pediatr 1995; 127: 298–300. 1984; 3: 407–409.