Scholarly article on topic 'Pediatric Within-Day Biological Variation and Quality Specifications for 38 Biochemical Markers in the CALIPER Cohort'

Pediatric Within-Day Biological Variation and Quality Specifications for 38 Biochemical Markers in the CALIPER Cohort Academic research paper on "History and archaeology"

0
0
Share paper
Academic journal
Clinical Chemistry
OECD Field of science
Keywords
{""}

Academic research paper on topic "Pediatric Within-Day Biological Variation and Quality Specifications for 38 Biochemical Markers in the CALIPER Cohort"

Clinical Chemistry 60:3 518-529 (2014)

Pediatric Clinical Chemistry

Pediatric Within-Day Biological Variation and Quality Specifications for 38 Biochemical Markers in the

CALIPER Cohort

Dana Bailey,1-2'3* Victoria Bevilacqua,1'2+ David A. Colantonio,1,2 Maria D. Pasic,1-2-4 Nandita Perumal,5

Man Khun Chan,1 and Khosrow Adeli1,2*

background: Studies of biological variation provide insight into the physiological changes that occur within and between study participants. Values obtained from such investigations are important for patient monitoring and for establishing quality specifications. In this study we evaluated the short-term biological variation of 38 chemistry, lipid, enzyme, and protein analytes in a pediatric population, assessed the effect of age partitions on interindividual variation, and compared the findings to adult values.

methods: Four plasma samples each were obtained within 8 h from 29 healthy children (45% males), age 4-18 years. Samples were stored at —80 °C and analyzed in 3 batches, with samples from 9-10 study participants per batch. Within-person and between-person biological variation values were established using nested ANOVA after exclusion of outliers by use of the Tukey outlier test. Analytical quality specifications were established with the Fraser method.

results: Biological variation coefficients and analytical goals were established for 38 analytes. Age partitioning was required for 6 analytes. Biological variation characteristics of 14 assays (37%) were distinct from adult values found in the Westgard database on biological variation. Biological variation characteristics were established for 2 previously unreported analytes, uncon-jugated bilirubin and soluble transferrin receptor.

conclusions: This study is the first to examine biological variation and to establish analytical quality specifications on the basis of biological variation for common assays in a pediatric population. These results provide

insight into pediatric physiology, are of use for reference change value calculations, clarify the appropriateness of reference interval use, and aid in the development of quality management strategies specific to pediatric laboratories.

© 2013 American Association for Clinical Chemistry

Biological variation is an important factor to be considered when interpreting laboratory test results in a clinical setting. Studies of biological variation provide insight into the physiological changes that occur within and between individuals for a given analyte. Although several studies have explored biological variation in adult populations, few studies have examined these attributes in pediatric samples. This information is crucial in result interpretation for 3 key reasons. First, within- and between-person biological variation can be used to establish reference change values (RCV),4 which provide the information required to determine if a change in concentration of a specific analyte qualifies as clinically significant (1). Establishing the usual amounts of observed analyte fluctuation by use of within- and between-person biological variation is central to long-term monitoring and follow-up of children. It is also an especially important consideration for analytes with cyclical rhythms for which the time ofcollection may affect the expected reference interval (2 ).

The second reason why this information is important is that data information on biological variation can be used to establish quality specifications (2) such as bias, precision, and total allowable error (TE). These

1 CALIPER program, Department of Pediatric Laboratory Medicine, The Hospital

for Sick Children, Toronto, Ontario, Canada; 2 Department of Laboratory Med-

icine and Pathobiology, University of Toronto, Toronto, Ontario, Canada; 3 current address: Gamma-Dynacare Medical Laboratories, London, Ontario, Canada;

4 current address: Department of Laboratory Medicine, St. Joseph's Health Centre, Toronto, Ontario, Canada; 5 Department of Pediatrics, The Hospital for Sick Children, Toronto, Ontario, Canada. * Address correspondence to this author at: Clinical Biochemistry, The Hospital for Sick Children, University of Toronto, Toronto, Ontario, M5G 1X8 Canada. Fax 416-813-6257; e-mail khosrow.adeli@sickkids.ca. f Dana Bailey and Victoria Bevilacqua contributed equally to the work, and both

should be considered as first authors.

Received August 8, 2013; accepted December 3, 2013. Previously published online at DOI: 10.1373/clinchem.2013.214312 4 Nonstandard abbreviations: RCV, reference change value; TE, total allowable error; CV,, within-subject/intraindividual biological CV; CVG, between-subject/ interindividual biological CV; CALIPER, Canadian Laboratory Initiative on Pae-diatric Reference Intervals; A1AT, a-1 antitrypsin; AGP, a-1 acid glycoprotein; C3, complement component 3; CRP, C-reactive protein; STfR, soluble transferrin receptor; ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CK, creatine kinase; GGT, y glutamyl transferase; HDL-C, HDL cholesterol; LDH, lactate dehydrogenase; CVdd, between-day CV; CVA, total CV; II, index of individuality.

specifications can have important implications for result interpretation. For example, in adult populations, changes in bias can affect how many low-risk individuals are diagnosed with diabetes (3 ) or how many individuals with cholesterol values within reference intervals are further investigated or sent for treatment (4). Although the same misclassifications are likely present in pediatric cohorts, data are lacking.

Although there are several ways to establish quality specifications, a consensus statement states that the use of biological variation is the preferred method (5). According to the hierarchy, a model based on biological variation is second only to one in which quality specifications are determined on the basis of the effect of analytical performance on specific clinical decision-making. However, the latter approach can be incredibly time-consuming and difficult (5) and therefore, biological variation represents both the most rigorous and the most efficient means by which to establish quality specifications.

Finally, biological variation can provide important insight into the usefulness ofreference intervals for different laboratory tests. Specifically, biological variation data can be used to generate the index of individuality— the ratio of within- to between-individual variation for a given analyte. Although a high index of individuality (>1.4) suggests that the use of reference intervals would be appropriate, a low index of individuality (<0.6) suggests that other tools such as RCVs should also be taken into consideration (6).

Clearly, the production of data on biological variation can generate a wealth of information not only regarding the physiological changes that occur within and between individuals, but also on what quality specifications are appropriate for a given test and how and when to use reference intervals. Although such data are evidently valuable, there is a clear lack of information on biological variation in pediatric populations. As such, in this study we aimed to evaluate the short-term (8 h) biological variation of 38 chemistry, lipid, enzyme, and protein analytes in a pediatric population of 29 healthy children age 4-18 years. The within- and between-person biological variation (CV: and CVG) for these analytes were estimated and the effects of age-specific partitions on between-person variation were assessed. Differences in biological variation between adult and pediatric populations were also evaluated. RCVs and indices of individuality were generated for each of the 38 analytes and, finally, the estimates ofbiological variation were used to generate analytical goals.

Materials and Methods

study participants

Healthy community children were recruited to take part in this 1-day (9:30-17:30) study via the Canadian

Laboratory Initiative on Paediatric Reference Intervals (CALIPER) outreach program (7). The study was completed at The Hospital for Sick Children in Toronto, Canada, with institutional ethics board approval. Before participation, the health of each child was confirmed via interview and by health and lifestyle questionnaires, with specific exclusion criteria as described (7). Children were instructed to fast overnight for at least 8 h before the study. A total of 30 children ages 4-18 years were recruited; 1 child was removed from the study owing to an inability to obtain sufficient samples at all time points. The demographics of the children have been previously reported (8 ). For each participant, blood was drawn at 4 time points, with a mean time period of approximately 2.5 h between each collection. Sample collection was performed after an overnight fast, midmorning after breakfast, within 2 h after lunch, and late afternoon. After collection into serum separator tubes, all samples were processed by the clinical chemistry laboratory and stored at —80 °C until batch testing.

analytical processing

To avoid multiple freeze-thaw cycles and to minimize analytical variability, all 38 assays were run on all samples from each child within the same day. Assays were completed in batches on the Cobas Integra 400 (Roche) [a-1 antitrypsin (A1AT), a-1 acid glycoprotein (AGP), complement component 3 (C3), C4, ceruloplasmin, C-reactive protein (CRP), haptoglobin, IgG, IgA, IgM, soluble transferrin receptor (STfR), transferrin] and the VITROS 5,1 FS chemistry systems analyzer (Ortho Clinical Diagnostics) [albumin, alkaline phosphatase (ALP), alanine aminotransferase (ALT), amylase, aspartate aminotransferase (AST), unconjugated bilirubin, calcium, cholesterol, creatine kinase (CK), chloride, CO2, creatinine, iron, y glutamyl transferase (GGT), glucose, HDL cholesterol (HDL-C), potassium, lactate dehydrogenase (LDH), magnesium, sodium, phosphate, total bilirubin, total protein, triglyceride, uric acid, urea] over a 3-day period, with samples from 9-10 individuals run per day. All assays were performed according to the manufacturers' recommendations. Before use, both analyzers were calibrated using manufacturer-provided reagents and calibrators. QC materials and 2 patient samples of known analyte concentrations were run with each batch. Analytical specifications for the 38 analytes tested have already been published (7).

statistical analysis

SPSS Statistical Software (version 21, IBM), EP Evalu-ator (version 9), and GraphPad Prism (version 4.0) were used for data analysis. Data from males and females were analyzed together. All data followed a

gaussian distribution (8). Outliers were eliminated by use of the Tukey method, which defines an outlier as 1.5 interquartile ranges below the 25th percentile or above the 75th percentile (9).

Within- and between-person biological variation were calculated using GraphPad Prism and STATA and expressed as CVs, CVI and CVG, respectively. CVG was estimated using a nested ANOVA. CVI was estimated by averaging the CV for each child, defined as the square root of the variance in results divided by the mean concentration for that individual, across all individuals. In contrast to estimates of CVI, for which the overall mean is used [e.g., (10)]. This approach is more appropriate for situations in which the overall mean is not representative of the mean for each individual. Both within- and between-person biological variation contain components of biological and analytical variation, which were minimized through experimental design and removed mathematically by subtracting the between-day CV (CVdd) or the total CV (CVA), respectively. Analytical CV was calculated using QC material run concurrently with the collected samples. The index of individuality (II), RCV, and analytical goals including imprecision, bias, and total allowable error, were calculated as follows:

II = CVI/CVG (note that the CVA was not included in

calculating II) (11); RCV = z(2)1/2(CVA + CV2)1/2; z = 1.96 for P < 0.05

for a bidirectional change; Analytical imprecision:

Optimal = 0.25 X CVI, Desirable = 0.50 X CVI, Minimal = 0.75 X CVI; Analytical bias:

Optimal = 0.125 X (CV2 + CVG)1/2, Desirable = 0.25 X (CV2 + CVG)1/2, Minimal = 0.375 X (CV2 + CVG)1/2; Total allowable error (TAE) = (1.65 X imprecision) + bias.

When statistically indicated, the population of children was partitioned according to age on the basis of previously determined CALIPER reference interval age partitions (7). To determine whether CALIPER-determined age partitions were appropriate for this substudy population, a Student i-test and F-test were performed on analytes containing 2 age partitions (uric acid, HDL-C, CRP, urea, ALT, haptoglobin, IgG), whereas a 1-way ANOVA and Bartlett test were performed on analytes containing 3 or more age partitions (total bilirubin, CO2, creatinine, phosphate, ALP, AST, LDH, albumin, total protein, and IgA) to

identify significantly different group means or variances, respectively. Reference intervals were not available for AGP, A1AT, bilirubin (unconjugated), ceruloplasmin, CK, chloride, glucose, potassium, sodium, and STfR (7).

Adult values for CVI and CVG were obtained from the Westgard Biological Variation database (12). This database is an updated compilation of multiple papers examining biological variation in adult populations (13). The weighted-mean CVG across pediatric age partitions was used for the pediatric cohort. Pediatric to adult CVI, CVG, RCV, and total allowable error (TE) ratios <0.5 or >1.5 were arbitrarily defined as being significantly different.

Results

A total of 29 children ages 4-18 years (13 male, 16 female) were recruited to participate in this 1-day (8 h) study. Before participation, the health ofeach child was ensured through an interview as well as health and lifestyle questionnaires. Participants were instructed to fast for at least 8 h before the study. Each child was sampled at 4 time points with approximately 2.5 h between each sampling, yielding 116 data points for each analyte; a total of 38 analytes were examined, yielding 4408 analytical results.

To correctly estimate CVG, it was necessary to partition children by age for 6 of the 38 analytes, specifically ALP, AST, creatinine, LDH, phosphate, and uric acid (Fig. 1, panel A). Suitable age partitions for this cohort were derived from the CALIPER database, as recently described (14). To determine whether age partitioning was necessary to minimize CVG, children were partitioned by age, and the mean concentration and variance of each age group was compared statistically, as described in Methods. Partitioning by age did not significantly affect CVG for HDL-C, CRP, albumin, total bilirubin, and IgG (Fig. 1B).

Table 1 lists the CVA, between-run CV, and CVdd CVs, the CVI and CVG, tAhe II, and the RCV for each oddf the 38 analytes, as partitioned by age. Only 1 analyte, AST, had an index of individuality that exceeded 1.4. The majority of analytes examined (albumin, ALP, AGP, A1AT, ALT, amylase, AST, C3, C4, ceruloplasmin, cholesterol, CK, chloride, CRP, iron, GGT, haptoglobin, HDL-C, IgA, IgG, IgM, LDH, STfR, total protein, transferrin, uric acid, and urea) showed marked individuality, with II values of <0.6.

When comparing the pediatric cohort and adult populations, CVI and CVG components of 24 of 38 analytes (63%) were found to be consistent with adult values published in the Westgard database (Table 2) (12). Four analytes showed marked differences, arbitrarily defined as a more than 50% reduction or more

Fig. 1. Relationship of within- and between-person biological variation with age in a pediatric cohort.

Numbers on the у axis refer to the study participant identification numbers, with children sorted by ascending age. The range of values for each child is indicated by the box-and-whisker plot. (A), Representative analytes for which within- and/or between-person biological variation change with age in a pediatric cohort. Dark grey shading indicates that the age partitions determined by prior CALIPER reference interval studies [Colantonio et al. (7)) significantly affect estimates of CV, and/or CVG. (B), Representative analytes for which within- and/or between-person biological variation do not change significantly with age in a pediatric cohort. Light grey shading indicates that the age partitions determined by prior CALIPER reference interval studies [Colantonio et al. (7)] did not significantly affect estimates of CV, and/or CVG.

than 150% increase, in both within- and between-person variation between the pediatric and adult populations. Specifically, CRP values in the pediatric cohort demonstrated reduced CV: and increased CVG compared to adult values, GGT showed reduced CV: and CVG, and ceruloplasmin and glucose showed increased CVj and CVG (Table 2). Additionally, 10 analytes showed marked differences in either CV: or CVG. Specifically, AGP, AST, cholesterol, CK, HDL, IgG, and STfR had reduced CVI, sodium had reduced CVG, and iron and transferrin had increased CVG. Interestingly, for iron, the smaller CVI and larger CVG values seen in the pediatric population resulted in a smaller II (0.38 vs 1.14) (12). As a consequence of the differences observed in CVI, the pediatric RCV was lower than that of

adult populations for AGP, AST, CK, CRP, GGT, and STfR, whereas it was increased for ceruloplasmin and glucose.

To provide a guide for analytical quality specifications for pediatric testing based on biological variation, we calculated optimal, desirable, and minimal analytical goals for imprecision, bias, and total allowable error (Table 3) and compared them with adult specifications (Table 2). The total allowable error based on biological variation characteristics was reduced by more than half of that allowed for adult populations for CK (14.3% vs 30.3%, pediatric vs adult, respectively) and STfR (6.8% vs 17.4%); it was increased to >150% of that allowed for an adult population for ceruloplasmin (15.1% vs 7.9%) and glucose (13.1% vs 7.2%).

Table 1. Biological variation characteristics in a pediatric cohort.

Analytical variation Biological variation

Analyte n Age, years Mean CVrr, % CVdd, % CVA, % CVi, % CVg, % II RCV, %

Albumin, g/dL 29 1 to <19 4.5 1.5 0.2 1.5 2.3 4.7 0.5 7.5

ALP, U/L 5 1 to <10 241.1 1.9 1.0 2.0 5.6 27.2 0.1 16.4

7 10 to <13 219.0 24.0 0.1

3 13 to <15 215.8 27.2 0.1

4 15 to <17 151.4 60.9 0.1

10 17 to <19 86.5 18.7 0.2

AGP, mg/dL 28 1 to < 19 80.0 0.9 0.7 1.1 3.7 20.5 0.2 10.8

A1AT, mg/dL 28 1 to < 19 120.0 1.9 0.0 1.9 5.0 13.0 0.4 14.7

ALT, U/L 29 1 to < 19 16.4 4.8 5.1 7.0 15.6 27.7 0.6 47.4

Amylase, U/L 29 1 to < 19 67.0 2.9 1.6 3.3 5.1 24.9 0.2 16.7

AST, U/L 2 1 to < 7 35.5 1.7 0.8 1.9 4.7 0.6 8.4 13.9

6 7 to <12 28.5 21.5 0.2

21 12 to <19 22.9 23.9 0.2

Bilirubin (total), mg/dL 29 1 to < 19 0.3 1.7 2.0 2.6 28.1 38.2 0.7 78.3

Bilirubin (unconjugated), mg/dL 28 1 to < 19 0.2 1.7 2.6 3.1 51.1 57.0 0.9 141.8

C3, mg/dL 29 1 to < 19 120.0 0.9 2.5 2.7 4.8 12.1 0.4 15.2

C4, mg/dL 24 1 to < 19 20.0 2.6 2.1 3.4 5.5 28.1 0.2 18.0

Calcium, mg/dL 29 1 to < 19 10.0 1.0 0.9 1.3 1.6 2.5 0.7 5.7

Ceruloplasmin, mg/dL 28 1 to < 19 19.6 3.8 1.3 4.1 11.3 20.3 0.6 33.4

Chloride, mmol/L 29 1 to < 19 105.9 0.2 0.5 0.6 0.8 1.5 0.6 2.8

Cholesterol, mg/dL 29 1 to < 19 166.0 0.9 1.1 1.5 2.4 15.7 0.2 7.9

CO2, mmol/L 29 1 to < 19 24.6 2.8 2.3 3.7 3.4 5.3 0.6 13.8

CK, U/L 28 1 to < 19 89.5 1.7 4.3 4.7 4.1 43.4 0.1 17.2

Creatinine, mg/dL 2 2 to < 5 0.3 1.1 0.7 1.3 4.2 4.1 1.1 12.3

5 5 to <12 0.5 23.3 0.2

8 12 to <15 0.6 11.0 0.4

14 15 to <19 0.8 15.7 0.3

CRP, mg/L 27 1 to < 19 0.9 1.6 2.9 3.3 19.3 125.4 0.2 54.1

GGT, U/L 28 1 to < 19 18.0 0.9 0.5 1.1 2.7 18.7 0.1 8.0

Glucose, mg/dL 29 1 to < 19 90.0 1.2 0.0 1.2 11.4 9.1 1.3 31.8

Haptoglobin, mg/dL 29 1 to < 19 90.0 1.2 1.3 1.7 10.7 50.8 0.2 30.1

HDL-C, mg/dL 25 1 to < 19 54.0 2.2 3.4 4.1 2.9 22.2 0.1 13.9

IgA, mg/dL 29 1 to < 19 150.0 0.6 1.3 1.5 4.4 42.1 0.1 12.9

IgG, mg/dL 29 1 to < 19 1110.0 2.4 1.1 2.6 1.1 14.7 0.1 7.8

IgM, mg/dL 28 1 to < 19 120.0 0.6 1.9 2.0 4.0 37.3 0.1 12.3

Iron, ^g/dL 29 1 to < 19 83.2 2.3 4.2 4.7 14.6 38.9 0.4 42.5

LDH, U/L 5 10 1 to < 10 10 to <15 652.9 528.4 2.0 0.9 2.2 4.7 12.5 18.1 0.4 0.3 14.4

14 15 to <19 420.2 13.0 0.4

Magnesium, mg/dL 29 1 to < 19 1.9 1.7 0.0 1.7 2.7 4.5 0.6 8.8

Continued on page 523

Table 1. Biological variation characteristics in a pediatric cohort. (Continued from page 522)

Analytical variation Biological variation

Analyte n Age, years Mean CVrr, % CVdd, % CVA, % CVI, % CVg, % II RCV, %

Phosphate, mg/dL 2 1 to <5 5.3 1.1 0.0 1.1 6.0 7.0 0.9 16.9

9 5 to <13 4.6 8.0 0.8

8 13 to <16 4.6 10.4 0.6

10 16to <19 4.0 5.1 1.2

Potassium, mmol/L 29 1 to <19 4.4 0.6 1.5 1.6 4.6 5.3 0.9 13.5

Sodium, mmol/L 29 1 to <19 142.7 0.5 0.6 0.8 0.4 0.4 0.9 2.4

STfR, mg/dL 27 1 to <19 350.0 3.2 0.0 3.2 1.4 22.4 0.1 9.6

Total protein, g/dL 29 1 to <19 8.0 1.5 0.8 1.7 1.7 4.6 0.4 6.7

Transferrin, mg/dL 28 1 to <19 2.2 1.9 0.0 1.9 3.0 10.8 0.3 9.8

Triglycerides, mg/dL 29 1 to <19 97.3 1.0 0.4 1.1 27.0 24.8 1.1 74.9

Urea, mg/dL 29 1 to <19 14.6 1.3 0.7 1.5 7.5 22.1 0.3 21.3

Uric acid, mg/dL 7 1 to <12 3.3 1.1 0.7 1.3 6.8 25.1 0.3 19.2

22 12 to <19 4.6 23.0 0.3

Discussion

Biological variation and its application in serial results monitoring and analytical goal establishment has been widely examined in adult populations. However, there is no literature available exploring biological variation in a pediatric population. As a consequence, pediatric laboratories have been obliged to adopt RCV and analytical quality goals based on adult cohorts. In the absence of data on biological variation, the clinical pediatric laboratory has been unable to provide a complete picture of the dynamic changes expected for analytes in the pediatric population. This is especially crucial given the many changes associated with childhood and teenage development. Previous CALIPER studies have highlighted the substantial differences between pediat-ric and adult reference intervals (7, 15, 16) and the effects of fasting and sampling time in a pediatric cohort (8). However, addressing this gap in information relevant to the pediatric population presents particular challenges, including the acquisition of samples and the necessary introduction of age partitioning to provide an accurate estimate of CVG. By examining the short-term biological variation of 38 chemistry analytes in a pediatric cohort, we have made the first steps to fill this gap in pediatric laboratory medicine. The age partitions established in previous CALIPER studies were tested in this analysis and it was determined that age partitioning was required for 6 of the 38 analytes examined, specifically, ALP, AST, creatinine, LDH, phosphate, and uric acid (Fig. 1A).

The CVI and CVG estimates in this pediatric population were found to be largely consistent with those

for adults. A reduced CVI was observed for 9 analytes, but this may have resulted from the short-term nature of this study. Previous studies have determined that measurements made 24 h apart had smaller CVI values than those taken at intervals of 4 days or longer (17). Additionally, an increased CVG for pediatric vs adult populations was observed for 5 analytes, which may indicate a need for further age and/or sex partitioning, as previously demonstrated by reference interval studies (18). Only 4 analytes, CRP, GGT, ceruloplasmin, and glucose, showed marked differences in both within- and between-person variation in the pediatric vs the adult populations (Table 2).

CRP presented with a reduced CVI (19.3% vs 42.2%) and a larger CVG (125.4% vs 76.3%) in the pediatric population. This increase in CVG largely resulted from 5 children with median CRP concentrations greater than approximately 1.5 mg/L (Fig. 2). Although the reason for the increase in CRP in these individuals remains unknown, the findings are consistent with NHANES (National Health and Nutrition Examination Survey) data for 10- to 15-year-old children, in which 15% of children had CRP values between 0.9 and2.7mg/Land 10% had values >2.7 mg/L (19).

GGT demonstrated a reduced CVI and CVG (Fig. 2). Because increases in GGT in adults are known to be nonspecific, with modest increases noted in conjunction with various common metabolic risk factors such as increased blood pressure and decreased HDL-C (20 ), the tighter biological control of GGT observed in children may be partially explained by an absence of these subclinical conditions.

Table 2. Comparison between pediatric and adult biological variation characteristics.

Pediatric

Adulta

Analyte N Age, years CV|, % CVG, %b RCV, % TE, %c CV|, % CVg, % RCV, % TE, %c

Albumin 29 1 to <19 2.3 4.7 7.5 3.2 3.1 4.2 9.5 3.9

ALP 29 1 to <19 5.6 28.1 16.4 11.8 6.4 24.8 18.2 11.7

AGP 28 1 to < 19 3.7d 20.5 10.8d 8.3 11.3 24.9 31.5 16.2

A1AT 28 1 to < 19 5.0 13.0 14.7 7.6 5.9 16.3 16.8 9.2

ALT 29 1 to < 19 15.6 27.7 47.4 20.8 18.0 42.0 53.5 26.3

Amylase 29 1 to < 19 5.1 24.9 16.7 10.6 8.7 28.3 25.8 14.6

AST 29 1 to < 19 4.7d 21.8 13.9d 9.5 11.9 17.9 33.4 15.2

Bilirubin (total) 29 1 to < 19 28.1 38.2 78.3 35.0 23.8 39.0 66.4 31.1

Bilirubin (unconjugated) 28 1 to < 19 51.1 57.0 141.8 61.3

C3 29 1 to < 19 4.8 12.1 15.2 7.2 5.2 15.6 16.2 8.4

C4 24 1 to < 19 5.5 28.1 18.0 11.7 8.9 33.4 26.4 16.0

Calcium 29 1 to < 19 1.6 2.5 5.7 2.1 1.9 2.8 6.4 2.4

Ceruloplasmin 28 1 to < 19 11.3e 20.3e 33.4e 15.1e 5.8 11.1 19.7 7.9

Chloride 29 1 to < 19 0.8 1.5 2.8 1.1 1.2 1.5 3.7 1.5

Cholesterol 29 1 to < 19 2.4d 15.7 7.9 6.0 5.4 15.2 15.5 8.5

CO2 29 1 to < 19 3.4 5.3 13.8 4.4 4.8 5.3 16.8 5.7

CK 28 1 to < 19 4.1d 43.4 17.2d 14.3d 22.8 40.0 64.5 30.3

Creatinine 29 2 to < 19 4.2 14.9 12.3 7.3 6.0 14.7 17.0 8.9

CRP 27 1 to < 19 19.3d 125.4e 54.1d 47.6 42.2 76.3 117.3 56.6

GGT 28 1 to < 19 2.7d 18.7d 8.0d 7.0d 13.8 41.0 38.4 22.2

Glucose 29 1 to < 19 11.4e 9.1e 31.8e 13.1e 6.1 6.1 17.1 7.2

Haptoglobin 29 1 to < 19 10.7 50.8 30.1 21.8 20.4 36.4 56.7 27.3

HDL-C 25 1 to < 19 2.9d 22.2 13.9 8.0 7.1 19.7 22.7 11.1

igA 29 1 to < 19 4.4 42.1 12.9 14.2 5.4 35.9 15.5 13.5

igG 29 1 to < 19 1.1e 14.7 7.8 4.6 4.5 16.5 14.4 8.0

igM 28 1 to < 19 4.0 37.3 12.3 12.7 5.9 47.3 17.3 16.8

Iron 29 1 to < 19 14.6 38.9e 42.5 22.4 26.5 23.2 74.6 30.7

LDH 29 1 to < 19 4.7 14.7 14.4 7.7 8.6 14.7 24.6 11.4

Magnesium 29 1 to < 19 2.7 4.5 8.8 3.5 3.6 6.4 10.7 4.8

Phosphate 29 1 to < 19 6.0 7.6 16.9 7.4 8.5 9.4 23.7 10.2

Potassium 29 1 to < 19 4.6 5.3 13.5 5.5 4.8 5.6 14.0 5.8

Sodium 29 1 to < 19 0.4 0.4d 2.4 0.5 0.7 1.0 2.9 0.9

STfR 27 1 to < 19 1.4d 22.4 9.6d 6.8d 13.6 20.8 38.4 17.4

Total protein 29 1 to < 19 1.7 4.6 6.7 2.6 2.7 4.0 8.8 3.4

Transferrin 28 1 to < 19 3.0 10.8e 9.8 5.3 3.0 4.3 9.1 3.8

Triglycerides 29 1 to < 19 27.0 24.8 74.9 31.4 20.9 37.2 58.0 27.9

Urea 29 1 to < 19 7.5 22.1 21.3 12.0 12.3 18.3 34.3 15.7

Uric acid 29 1 to < 19 6.8 23.5 19.2 11.7 9.0 17.6 25.2 12.4

a Westgard website available: http://www.westgard.com/blodatabase1.htm. b Weighted CVG.

c Desirable TE = 1.65 (0.5 X CV,) + [0.25 X /(CV,2 + CVG2)]. d Reduced variation compared to adult cohorts (pediatric CV/adult CV <0.5). e Increased variation compared to adult cohorts (pediatric CV/adult CV >1.5).

Table 3. Analytical goals for pediatric testing based on short-term biological variation.

Analytical goals

Imprecision, CV, % Bias, % TE, %

Analyte Age, Years Optimal Desirable Minimal Optimal Desirable Minimal Optimal Desirable Minimal

Albumin 1 to <19 0.6 1.1 1.7 0.6 1.3 1.9 1.6 3.2 4.8

ALP 1 to <10 1.4 2.8 4.2 3.5 6.9 10.4 5.8 11.5 17.3

10 to <13 3.1 6.2 9.2

13 to <15 3.5 7.0 10.4

15 to <17 7.6 15.3 22.9

17 to <19 2.4 4.9 7.3

AGP 1 to < 19 0.9 1.9 2.8 2.6 5.2 7.8 4.1 8.3 12.4

A1AT 1 to < 19 1.2 2.5 3.7 1.7 3.5 5.2 3.8 7.6 11.3

ALT 1 to < 19 3.9 7.8 11.7 4.0 8.0 11.9 10.4 20.8 31.2

Amylase 1 to < 19 1.3 2.5 3.8 3.2 6.3 9.5 5.3 10.5 15.8

AST 1 to < 7 1.2 2.3 3.5 0.6 1.2 1.8 2.5 5.0 7.5

7 to <12 2.7 5.5 8.2

12 to <19 3.0 6.1 9.1

Bilirubin (total) 1 to < 19 7.0 14.1 21.1 5.9 11.9 17.8 17.5 35.1 52.6

Bilirubin (unconjugated) 1 to < 19 12.8 25.6 38.3 9.6 19.1 28.7 30.7 61.3 92.0

C3 1 to < 19 1.2 2.4 3.6 1.6 3.2 4.9 3.6 7.2 10.8

C4 1 to < 19 1.4 2.8 4.2 3.6 7.2 10.8 5.9 11.7 17.6

Calcium 1 to < 19 0.4 0.8 1.2 0.4 0.7 1.1 1.0 2.1 3.1

Ceruloplasmin 1 to < 19 2.8 5.7 8.5 2.9 5.8 8.7 7.6 15.2 22.8

Chloride 1 to < 19 0.2 0.4 0.6 0.2 0.4 0.6 0.5 1.1 1.6

Cholesterol 1 to < 19 0.6 1.2 1.8 2.0 4.0 6.0 3.0 6.0 8.9

CO2 1 to <19 0.8 1.7 2.5 0.8 1.6 2.3 2.2 4.3 6.5

CK 1 to < 19 1.0 2.0 3.0 5.4 10.9 16.3 7.1 14.2 21.4

Creatinine 2 to <5 1.1 2.1 3.2 0.7 1.5 2.2 2.5 5.0 7.5

5 to <12 3.0 5.9 8.9

12 to <15 1.5 3.0 4.4

15 to <19 2.0 4.1 6.1

CRP 1 to < 19 4.8 9.6 14.4 15.9 31.7 47.6 23.8 47.6 71.4

GGT 1 to < 19 0.7 1.3 2.0 2.4 4.7 7.1 3.5 6.9 10.4

Glucose 1 to < 19 2.9 5.7 8.6 1.8 3.7 5.5 6.5 13.1 19.6

Haptoglobin 1 to < 19 2.7 5.4 8.0 6.5 13.0 19.5 10.9 21.8 32.7

HDL-C 1 to < 19 0.7 1.4 2.2 2.8 5.6 8.4 4.0 8.0 11.9

IgA 1 to < 19 1.1 2.2 3.3 5.3 10.6 15.9 7.1 14.2 21.3

IgG 1 to < 19 0.3 0.6 0.8 1.8 3.7 5.5 2.3 4.6 6.9

igM 1 to < 19 1.0 2.0 3 4.7 9.4 14.1 6.3 12.7 19.0

Iron 1 to < 19 3.6 7.3 10.9 5.2 10.4 15.6 11.2 22.4 33.7

LDH 1 to < 10 1.2 2.4 3.5 1.7 3.3 5.0 3.6 7.2 10.8

10 to <15 2.3 4.7 7.0

15 to <19 1.7 3.5 5.2

Magnesium 1 to < 19 0.7 1.3 2.0 0.7 1.3 2.0 1.8 3.5 5.3

Continued on page 526

Table 3. Analytical goals for pediatric testing based on short-term biological variation. (Continued from

page 525)

Analytical goals

Imprecision, CV, % Bias, % TE, %

Analyte Age, Years Optimal Desirable Minimal Optimal Desirable Minimal Optimal Desirable Minimal

Phosphate 1 to <19 1.5 3.0 4.5 1.1 2.3 3.4 3.6 7.2 10.9

5 to <13 1.3 2.5 3.8

13 to <16 1.5 3.0 4.5

16to <19 1.0 2.0 2.9

Potassium 1 to < 19 1.1 2.3 3.4 0.9 1.7 2.6 2.8 5.5 8.3

Sodium 1 to < 19 0.1 0.2 0.3 0.1 0.1 0.2 0.2 0.4 0.7

STfR 1 to < 19 0.3 0.7 1.0 2.8 5.6 8.4 3.4 6.7 10.1

Total protein 1 to < 19 0.4 0.9 1.3 0.6 1.2 1.8 1.3 2.6 4.0

Transferrin 1 to < 19 0.7 1.5 2.2 1.4 2.8 4.2 2.6 5.3 7.9

Triglycerides 1 to < 19 6.8 13.5 20.3 4.6 9.2 13.7 15.7 31.5 47.2

Urea, 1 to < 19 1.9 3.8 5.6 2.9 5.8 8.7 6.0 12.0 18.1

Uric acid 1 to <12 1.7 3.4 5.1 3.2 6.5 9.7 6.1 12.1 18.2

12 to <19 3.0 6.0 9.0

Interestingly, glucose presented with increases in both CVI and CVG relative to adult populations (Fig. 2). These increases are believed to be due to the more prominent effect of fasting on glucose homeostasis in children. In comparison to adults, children are more prone to decreases in plasma glucose and increases in ketone body production [reviewed in (21 J], thereby widening the range of observed glucose concentrations. Indeed, 7 pediatric study participants had fasting glucose concentrations <4.0 mmol/L or 72 mg/dL, with a nonfasting upper limit of 8.0 mmol/L or 144 mg/dL.

Due to a decrease in CVI (14.6% vs 26.5%) and a modest increase in CVG (38.9% vs 23.2%), the II of iron was reduced from 1.14 to 0.38 (adult vs pediatric) (Fig. 2). Consistent with the reduction in CVI, it has previously been shown that infants and young children lack the diurnal variation of serum iron due to the absence of a sustained period of sleep (22 ). Furthermore, the discrepancy in CVG may be explained, in part, by the fact that iron deficiency is common in the pediatric population, with females aged 12-19 years at especially high risk for anemia (23). The observation that the CVG for a related analyte, transferrin, was also markedly increased in the pediatric population (10.8% vs 4.3%) suggests substantial variations in pediatric individuals with respect to iron status. Future long-term studies will be needed to explore whether or not this observation persists when the duration ofthe sampling period is increased.

The need for properly established analytical goals, and the consequences that result from a lack thereof, have been well documented (3, 4, 6, 24). As such, it was necessary to determine whether any differences exist in the laboratory test quality specifications required in the pediatric population compared with those in the adult population. Our analysis revealed that 4 analytes showed discernible differences in TE between the adult and pediatric populations: CK and STfR in the pediat-ric population had a decreased TE compared with the adult population, whereas ceruloplasmin and glucose had an increased TE (Table 2). It should be noted that interpretation of these results should take into consideration the clinical context and the downstream effect of clinical misclassification. For example, as a consequence of the increase in both components of biological variation, the calculated TE for glucose increased from 7.2% to 13.1%. However, we argue that the observation of increased susceptibility to hypoglycemia in pediatric individuals suggests the need for accurate and precise glucose measurements in this population, particularly at low concentrations.

In terms of the II, it has been argued that analytes with II values of <0.6 show a high degree of individuality and, therefore, the RCV as opposed to a reference interval should be used to assess the patient. On the other hand, analytes with II values of >1.4 show very little individuality and, therefore, the use of reference intervals is deemed to be appropriate (6). Analysis of the samples obtained from this pediatric population

Fig. 2. Within- and between-person biological variation characteristics unique to pediatric samples.

Estimates of CVG for glucose, GGT, CRP, and iron are indicated by the top horizontal line and vertical dashed lines. Numbers on the y axis refer to study participant identification numbers, with individuals sorted by ascending age. The range of values for each child is indicated by the box-and-whisker plot.

revealed that only 1 analyte, AST, had an II value that exceeded 1.4, whereas 27 of the 38 analytes examined had an II value below 0.6. Of particular interest, the reduction of II for iron to 0.38 suggests further investigation of the utility of an iron reference interval for clinical decision-making in a pediatric population.

Although it maybe appropriate to further examine the usefulness of population-based reference intervals in cases in which the II falls below 0.6, it is also important to consider the clinical context in which the ana-lyte is likely to be used (25). In many cases, a pathological state would result in a dramatic increase in analyte

concentration that would clearly fall outside of the established reference interval. For example, bilirubin has an II value of 0.74, which falls only slightly above the 0.6 cutoff, showing a fair degree of individuality. Reference intervals derived from a sample of healthy children indicate that total bilirubin for children of ages birth to 14 days should fall between 0.19 and 16.60 mg/dL, and for children ages 15 days to 1 year, total bilirubin should fall between 0.05 and 0.68 mg/dL (7). Cases of kernicterus have rarely been reported in neonates with bilirubin concentrations of <25 mg/dL and are not reported in neonates with bilirubin peak con-

centrations of <20 mg/dL (26). These clinically relevant values far exceed the reference intervals established, and it is unlikely that, despite the small II value for bilirubin, a clinical diagnosis based on bilirubin concentrations would be adversely affected by comparison with a population-based reference interval. Therefore, it is important that measures like II, as well as clinical context, be taken into account when interpreting results and determining a course of action for the patient in question.

Finally, it is important to note certain limitations of our study. First, previous CALIPER studies have determined that, for the creation of reference intervals, certain analytes (e.g., creatinine, albumin, and ALT) show sex differences and, therefore, must be partitioned accordingly (7, 16, 27). However, due to sample size limitations, we were unable to assess the appropriateness of partitioning CVI and CVG by sex. Larger-scale studies will be needed to determine the role of sex in biological variation for these analytes. Second, owing to the sample size of our study and the need to run samples in batches over several days, our protocol introduced components of between-run analytical imprecision into estimates of CVG which were removed mathematically. Third, owing to the challenges associated with obtaining pediatric samples, this study was restricted to a single day. Therefore, calculations such as analytical quality values or RCV might be relevant only for repeat samples collected within the same timeframe. Fourth, in comparing the pediatric estimates of CVI to those of adult populations, it was assumed that the within-person biological variation in a single day would be representative of long-term variation. Although we attempted to account for some effects of diurnal variation and fed/fasting-related variation by sampling in a fasting and postprandial state, the esti-

mates of CVI are likely smaller than what would be obtained from a study of longer duration. Lastly, given the short-term nature of this study, preanalytical sampling variation may contribute significantly to the variation observed. Future long-term studies will be needed to confirm the findings described herein.

In conclusion, this is the first study to analyze the components of biological variation in a healthy pediatric cohort. We have derived estimates of RCVs and quality indicators for precision, bias, and total allowable error. Future studies are needed to refine our estimates of the components of biological variation in pe-diatric populations, to extend our analysis of short-term biological variation to longer time periods, and to increase the sample size investigated.

Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.

Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:

Employment or Leadership: None declared. Consultant or Advisory Role: None declared. Stock Ownership: None declared. Honoraria: None declared. Research Funding: K. Adeli, CIHR. Expert Testimony: None declared. Patents: None declared.

Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled participants, review and interpretation of data, or preparation or approval of manuscript.

References

1. Erden G, Barazi AO, Tezcan G, Yildirimkaya MM. Biological variation and reference change values of CA 19-9, CEA, AFP in serum of healthy individuals. Scand J Clin Lab Invest 2008;68:212-8.

2. Fraser CG. Data on biological variation: essential prerequisites for introducing new procedures? Clin Chem 1994;40:1671-3.

3. Petersen PH, Jorgensen LG, Brandslund I, De Fine Olivarius N, Stahl M. Consequences of bias and imprecision in measurements of glucose and hba1c for the diagnosis and prognosis of diabetes mellitus. Scand J Clin Lab Invest Suppl 2005;240: 51-60.

4. Petersen PH, de Verdier CH, Groth T, Fraser CG, Blaabjerg O, Horder M. The influence of analytical bias on diagnostic misclassifications. Clin Chim Acta 1997;260:189-206.

5. Fraser CG, Kallner A, Kenny D, Petersen PH. Introduction: strategies to set global quality specifications in laboratory medicine. Scand J Clin Lab Invest 1999;59:477-8.

6. Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989;27:409-37.

7. Colantonio DA, Kyriakopoulou L, Chan MK, Daly CH, Brinc D, Venner AA, et al. Closing the gaps in pediatric laboratory reference intervals: a CALIPER database of 40 biochemical markers in a healthy and multiethnic population of children. Clin Chem 2012;58:854-68.

8. Pasic MD, Colantonio DA, Chan MK, Venner AA, Brinc D, Adeli K. Influence of fasting and sample collection time on 38 biochemical markers in healthy children: a CALIPER substudy. Clin Biochem 2012;45:1125-30.

9. Tukey J. Exploratory data analysis. Reading: Addison-Wesley Publishing Co; 1977.

10. Widjaja A, Morris RJ, Levy JC, Frayn KN, Manley SE, Turner RC. Within- and between-subject variation in commonly measured anthropometric and biochemical variables. Clin Chem 1999;45: 561-6.

11. Petersen PH, Fraser CG, Sandberg S, Goldschmidt H. The index of individuality is often a misinterpreted quantity characteristic. Clin Chem Lab Med 1999;37:655-61.

12. Westgard QC. Desirable specifications for total error, imprecision, and bias, derived from intra-and inter-individual biological variation. http:// www.westgard.com/biodatabase1.htm (Accessed June 2013).

13. Ricos C, Alvarez V, Cava F, Garcia-Lario JV, Hernandez A, Jimenez CV, et al. Current databases on biological variation: pros, cons and progress. Scand J Clin Lab Invest 1999;59:491-500.

14. Brinc D, Chan MK, Venner AA, Pasic MD, Colantonio D, Kyriakopolou L, Adeli K. Long-term stability of biochemical markers in pediatric serum specimens stored at -80 degrees C: a CALIPER Substudy. Clin Biochem 2012;45:816-26.

15. Bailey D, Colantonio D, Kyriakopoulou L, Cohen AH, Chan MK, Armbruster D, Adeli K. Marked biological variance in endocrine and biochemical

markers in childhood: establishment of pediatric reference intervals using healthy community children from the CALIPER cohort. Clin Chem 2013; 59:1393-405.

16. Konforte D, Shea JL, Kyriakopoulou L, Colantonio D, Cohen AH, Shaw J, et al. Complex biological pattern of fertility hormones in children and adolescents: a study of healthy children from the CALIPER cohort and establishment of pediatric reference intervals. Clin Chem 2013;59:1215-27.

17. Rotterdam EP, Katan MB, Knuiman JT. Importance of time interval between repeated measurements of total or high-density lipoprotein cholesterol when estimating an individual's baseline concentrations. Clin Chem 1987;33:1913-5.

18. Clifford SM, Bunker AM, Jacobsen JR, Roberts WL. Age and gender specific pediatric reference intervals for aldolase, amylase, cerulo-plasmin, creatine kinase, pancreatic amylase, prealbumin, and uric acid. Clin Chim Acta

2011;412:788-90.

19. Ford ES, Giles WH, Myers GL, Rifai N, Ridker PM, Mannino DM. C-reactive protein concentration distribution among US children and young adults: findings from the National Health and Nutrition Examination Survey, 1999-2000. Clin Chem 2003;49:1353-7.

20. Grundy SM. Gamma-glutamyl transferase: another biomarker for metabolic syndrome and cardiovascular risk. Arterioscler Thromb Vasc Biol 2007;27:4-7.

21. Lamers KJ, Doesburg WH, Gabreels FJ, Romsom AC, Renier WO, Wevers RA, Lemmens WA. Reference values of blood components related to fuel metabolism in children after an overnight fast. Clin Chim Acta 1985;145:17-26.

22. Schwartz E, Baehner RL. Diurnal variation of serum iron in infants and children. Acta Paediatr Scand 1968;57:433-5.

23. Cooper M, Greene-Finestone L, Lowell H,

Levesque J, Robinson S. Iron sufficiency of Canadians. Health Rep 2012;23:41-8.

24. Viljoen A, Singh DK, Farrington K, Twomey PJ. Analytical quality goals for 25-vitamin D based on biological variation. J Clin Lab Anal 2011;25: 130-3.

25. Omar F, van der Watt GF, Pillay TS. Reference change values: how useful are they? J Clin Pathol 2008;61:426-7.

26. Bhutani VK, Vilms RJ, Hamerman-Johnson L. Universal bilirubin screening for severe neonatal hyperbilirubinemia. J Perinatal 2010; 30(Suppl):S6-15.

27. Chan MK, Seiden-Long I, Aytekin M, Quinn F, Ravalico T, Ambruster D, Adeli K. Canadian Laboratory Initiative on Pediatric Reference Interval Database (CALIPER): pediatric reference intervals for an integrated clinical chemistry and immunoassay analyzer, Abbott ARCHITECT ci8200. Clin Biochem 2009;42:885-91.