Capillary blood stability and analytical accuracy of 12 analytes stored in Microtainers®

Abstract

Objectives

The aim of this study was to determine feasibility of collecting capillary blood by traditional fingerstick and next day analysis after transport in Microtainers® at ambient temperature with no plasma separation. This study is pursuing an acceptable alternative to venipuncture for measuring 12 analytes important for health risk assessment.

Design

and Methods: Performance standards of a 12-assay chemistry panel were assessed using a set of paralleled serum and capillary microsamples. The panel included Hemoglobin A1c (HbA1c), Total Cholesterol, Triglycerides, HDL-C, Creatinine, Urea Nitrogen (BUN), Uric Acid, alkaline phosphatase (ALP), ALT (GPT), AST (GOT), gamma-glutamyltransferase (GGT), and total protein. Correlation studies were performed using 31 simultaneous venous and capillary blood collections. Analytical bias, correlation, and medical decision points were calculated to determine equivalency of sample type and the impact of transport conditions. Clinical sensitivity, specificity, and predictive values were evaluated at calculated medical decision points for their usability in health screening initiatives.

Results

Laboratory test results using capillary blood samples stored in Microtainers® under conditions of delayed centrifugation, and mail transport at ambient temperature, showed an acceptable agreement with results obtained using their paired serum samples analyzed using standard methods, except AST.

Conclusions

Capillary blood samples can be self-collected at remote locations using Microtainers® and transported at ambient temperature for 24 h for successful performance of several medical tests important in large-scale health screenings programs.

Highlights

• •We evaluated a clinical chemistry panel of analytes using capillary blood collected and stored in Microtainers®.
• •Samples are stable at ambient temperature for 24 h, allowing for transportation by express mail.
• •The method favorably compares to blood samples obtained using standard venipuncture methods.

1. Introduction

While most diseases remain silent during their subclinical phase, the biochemistry of human blood is typically altered providing early signals of an evolving pathology. Changes are usually detected by common laboratory tests using blood specimens obtained by a medical technician [1]. However, this practice represents a logistic barrier for disease screening purposes. It requires travel to a patient service center, or healthcare workers must travel to multiple locations to collect and transport blood to a central laboratory for analysis. This approach is becoming increasingly unnecessary due to point-of-care alternatives that require small volumes of capillary blood, but their use remains limited by higher costs and availability of new tests [2]. Clinical analysis of a small amount of blood is now also possible in modern laboratory instruments. For example, a single drop of blood can be used to obtain a complete blood cell count, or Hemoglobin A1c [3,4]. Microtubes have also been used to collect and transport blood samples obtained by a finger skin puncture. Microtubes can also be used while attached to novel vacuum-powered devices such as those manufactured by Tasso, and YourBioHealth allowing faster and easier capillary microsample collection [[5], [6], [7]]. Microsamples have been traditionally collected using a dried blood spot modality due to its greater stability but limited by hemoglobin interference in photometric analyzers. Therefore, whole blood microsampling has been recently seen as an alternative despite preanalytical requirements on centrifugation and refrigerated transport to the laboratory. Since self-collected “home-based” blood samples can be neither centrifuged nor a cold chain maintained, it becomes important to determine the stability of selected analytes without centrifugation and transport at higher temperature conditions. Previous studies of capillary blood microsamples have shown good agreement between analyte concentration in serum and capillary blood processed the same day [8]. However, studies showing whether unprocessed samples and kept at ambient temperature are suitable for delayed analysis are not available. This study is aimed at the feasibility of performing select clinical laboratory tests on blood microsamples collected and transported in mint top Microtainers® under field conditions.

2. Materials and methods

2.1. Clinical specimens

Paired venous and capillary blood samples from 31 apparently healthy adult donors (14 females, 17 males) were collected for comparison studies. Venous samples were collected according to standard phlebotomy procedures using serum separator and EDTA vacutainer® tubes (BD # 367986, 367835). Serum samples were centrifuged for 15 min at 2,000×g prior to transportation under refrigerated conditions using a styrofoam container with cold packs according to standard pre-analytic good laboratory practices. All venous samples were de-capped and loaded in the analyzer upon arrival to the laboratory. Capillary blood was obtained by fingerstick using a disposable contact-activated lancet (BD # 366594), and then collected into a lithium heparin (mint top) Microtainer® (BD # 365985). Immediately after the venous draw was performed, the phlebotomist supervised each donor through the microtainer self-collection process. The skin puncture method has been described elsewhere [8]. Microsamples (ca. 0.3 mL each) were mixed by inverting the sample 10 times to prevent clotting and transported uncentrifuged and without refrigeration. All specimens were shipped by express mail using a UN3373 label according to federal and international regulations for the transport of a biological substance category B. Before analysis, each microsample was mixed once again and a 0.01-mL aliquot was retrieved for Hemoglobin A1c testing. The microsamples were then centrifuged at 3,000×g for 5 min, the plasma transferred into Hitachi sample cups (Roche # 10394246001) using disposable narrow stem transfer pipettes and analyzed within 1 h to avoid variability due to evaporation. Most blood specimens were collected during summer in four small separate groups, shipped from different locations, and monitored for heat exposure during transit using USB TE-02 PRO (Aprvtio) temperature data recorders. Drop boxes located on sidewalks and shipping over weekend days were avoided. The entire study was conducted using anonymized samples from selected participants at health screening events. Informed consent for the study was obtained and approved by the ethics committee at CoreMedica Laboratories in accordance with the Declaration of Helsinki [9].

2.2. Analytical instrument

All analyses were performed on a Roche c502 COBAS 8000 modular platform with manufacturer's reagents and protocols [10]. Sample interference measurements were obtained for hemoglobin (H-index) to assess hemolysis in each sample. Analytes routinely analyzed in our laboratory were included in the study. Internal quality control samples were evaluated every 4 h for all analytes and were monitored by external quality assurance programs.

2.3. Method comparison studies

Two approaches were used to study comparability between venous samples, considered the “gold standard”, and capillary blood samples, collected and transported in BD microtubes. Deming regression was used to determine the relationship of the medical decision point (MDP) for each analyte in both methods. Regression analyses were performed using the EP evaluator software (Data Innovations, Colchester, VT). The agreement of microtube results with regular venous samples was evaluated using Bland-Altman difference graphs between individual pair values (venous, capillary) on the vertical axis against to the average values of the individual pair values using a limit of agreement set at mean ± 1.96 times the standard deviation (SD) of the differences [11]. Difference plots were analyzed using Microsoft Excel.

2.4. Evaluation of total error and microsample method effectiveness

Paralleled clinical tests in the cohort of 31 donors were counted for the number of microsamples outside the published total allowable error for venous samples [12]. Microsamples were also evaluated for effectiveness based on clinical sensitivity, specificity, and predictive value calculations from paralleled assay results. Assay results were considered either positive or negative using a cut-off value representing the medical decision point in each case. Calculations were made comparing venous and capillary blood results side-by-side and counting the number of true positives (a), false positives (b), false negatives (c), and true negatives (d) to determine sensitivity (a/a+c), specificity (d/b + d), positive predictive value (a/a+b), and Negative predictive value (d/c + d) [13]. These four indices are calculated using venous samples as the reference to only evaluate the ability of microsamples to achieve equivalent health screening outcomes.

3. Results

3.1. Temperature range of clinical specimens

The remote collection of specimens in the health and wellness industry creates many opportunities for suboptimal specimen quality and mainly due to higher transit temperature. The temperature of samples after collection and during transport and its effect on hemolysis are shown in Table 1. Refrigerated venous samples used as reference laboratory values showed temperature fluctuations during pre-analytic processing and transport, with a mean value of 10 ± 3 °C consistent with standard protocols for shipping refrigerated specimens. Capillary blood microsamples showed a mean temperature between 20 and 25 °C and variability not exceeding 30 °C. Hemolytic interference in most COBAS assays is expected at hemoglobin index values greater than 200 mg/dL, except AST with a hemoglobin index limit at 60 mg/dL.

Table 1.

Specimen pre-analytic transit temperature and effect on hemolysis.

Blood Sample	Temperature Conditions	Mean Temperature (oC)	Range (oC)	Mean Hemoglobin Index (mg/dL)
Venous	Refrigerated	12.9	11.3–17.1	10
Capillary	Ambient	24.8	19.6–28.5	70

3.2. Comparability of results

Fig. 1 (Panels A through L) depicts comparison studies for each assay performed with conventional venous and paralleled capillary samples. Supporting statistics are shown in Table 2. An observed correlation over 97% demonstrates a good association between analyte measurements made in both alternative sample matrices. Using this criterion, both methods are statistically identical by sharing a slope of 1, and an intercept of 0 within 95% of their confidence interval for each analyte, despite lower correlation observed for uric acid and total protein.

Fig. 1.

Comparison of venous with capillary microsamples analyzed by Deming regression. Scatter plots bounds indicate 95% confidence interval limits.

Table 2.

Summary of regression statistics.

Analyte (Units)	a Slope (95% CI)	a Intercept (95% CI)	Correlation	Bias (%)
HbA1c (%)	0.977 (0.894 – 1.060)	-0.010 (-0.469 – 0.449)	0.9750	- 0.136 (-2.47 %)
Cholesterol (mg/dL)	0.997 (0.907 – 1.047)	9.8 (-4.0 – 23.6)	0.9822	5.3 (2.7 %)
Triglycerides (mg/dL)	0.963 (0.908 – 1.019)	4.4 (-5.4 – 14.1)	0.9885	- 1.1 (- 0.7 %)
HDL-C (mg/dL)	0.972 (0.897 – 1.046)	4.1 (-0.1 – 8.4)	0.9796	2.6 (4.7 %)
Creatinine (mg/dL)	0.955 (0.889 – 1.021)	- 0.015 (-0.079 – 0.049)	0.9836	- 0.057 (-6.1 %)
Urea Nitrogen (mg/dL)	1.048 (0.987 – 1.109)	0.2 (-0.7 – 1.1)	0.9884	0.9 (6.0 %)
Uric Acid (mg/dL)	0.932 (0.794 – 1.071)	- 0.20 (-0.95 – 0.55)	0.9240	- 0.55 (-10.5 %)
ALP (mg/dL)	0.981 (0.904 – 1.058)	- 1.7 (-8.2 – 4.7)	0.9795	- 3.3 (- 4.1%)
ALT (IU/L)	1.080 (1.040 – 1.120)	0.7 (-0.6 – 1.9)	0.9953	2.7 (10.6 %)
AST (mg/dL)	1.087 (0.982 – 1.193)	7.5 (4.4 – 10.7)	0.9674	9.6 (40.0 %)
GGT (mg/dL)	1.039 (0.974 – 1.104)	1.6 (0.0 – 3.3)	0.9865	2.5 (11.0 %)
Protein, Total (g/dL)	1.228 (0.925 – 1.530)	- 1.29 (-3.46 – 0.88)	0.7813	0.35 (4.8 %)

^a95% Confidence intervals are shown between parentheses.

Correlations over 90% allowed to the comparison of medical decision points (MDP) for each analyte. If both methods are identical, the 95% confidence interval for each calculated microsample MDP includes the corresponding MDP value established for venous samples. Table 3 shows microsamples as identical to conventional venous assays, except for HDL-C, uric acid, and liver enzymes. Total protein MDP comparison could not be calculated due to lower correlation under 90%.

Table 3.

Medical decision point analysis.

Analyte (Units)	Venous MDP	a Microsample MDP	b 95% CI Limits
HbA1c (%)	6.5	6.4	6.3–6.4
Cholesterol, Total (mg/dL)	240	244	240–249
Triglycerides (mg/dL)	200	197	191–203
HDL-C (mg/dL)	50	52	52–54
Creatinine (mg/dL)	1.4	1.3	1.3–1.4
Urea Nitrogen (mg/dL)	24	25	24–26
Uric Acid (mg/dL)	8.0	7.3	6.9–7.7
ALP (IU/L)	115	111	108–114
ALT (GPT) (IU/L)	55	60	59–62
AST (GOT) (IU/L)	48	60	57–63
GGT (IU/L)	60	64	61–66
Protein, Total (g/dL)	6.3	N/A	N/A

^aCalculated medical decision points for each analyte using microsamples.

^bConfidence interval for each analyte MDP in microsamples.

The capillary microsample data shows an assay bias of 0.7–11%, except for AST with a higher positive bias of 40%, despite the good linear relationship with its venous counterpart. Fig. 2 (Panels A through L) shows a significant agreement of analytical results between venous (“gold standard”) and blood microsamples when evaluated using Bland-Altman difference plots. Data analyses reveled no systematic biases and outliers in each case. These results indicate that differences are not clinically important, except for AST where most data points fall outside the limits of agreement (Fig. 2, Panel J).

Fig. 2.

Bland-Altman difference Plot of analytes in venous serum and plasma microsamples. Dashed lines represent the 95% confidence interval of the observed mean difference in each case.

3.3. Error count, sensitivity, specificity and predictive values

The total number of microsamples for each assay falling outside the acceptable error is shown in Table 4. The data shows a small error in each assay, except for AST where 26 out 31 microsamples reveal poor accuracy. The impact of error in microsample assay effectiveness, using reference MDP venous values (Table 3), is represented by shifts in clinical sensitivity, specificity, and predictive values shown in Table 4. All four indices indicate good equivalency between venous samples and capillary microsamples. High negative predictive values, except for triglycerides, provides evidence to consider both sampling collection methods as significantly equivalent.

Table 4.

Clinical sensitivity, specificity, and microanalysis error.

Analyte	aTAE (%)	Failed Samples	b Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
HbA1c	6	1	100	100	100	100
Cholesterol, Total	10	1	100	96	86	100
Triglycerides	20	3	78	100	100	92
HDL-C	15	2	100	85	90	100
Creatinine	15	1	100	100	100	100
Urea Nitrogen	9	3	100	100	100	100
Uric Acid	17	4	100	100	100	100
ALP	30	0	100	100	100	100
ALT (GPT)	20	3	100	97	67	100
AST (GOT)	20	26	100	97	50	100
GGT	20	4	100	100	100	100
Protein, Total	10	0	100	100	100	100

^aTotal allowable error values were defined using published government guidelines.

^bPPV, positive predictive value; NPV, negative predictive value.

4. Discussion

Clinical laboratory data from apparently healthy individuals have been shown to be helpful to detect silent medical conditions during pre-clinical stages [14]. Employers have used this strategy for many years to control health insurance costs and maintain productivity by reducing absenteeism [15]. Likewise, consumers use the same strategy to monitor therapeutic treatment or wellness goals [16]. These initiatives often require remote collection of a blood sample performed under limited supervision creating opportunities for suboptimal quality of laboratory services since samples must travel uncentrifuged for many hours at ambient temperature. Our laboratory has successfully used Microtainers® to reproduce microtube studies from different research groups, but their usability for routine chemistry analytes under remote blood collection conditions is unknown [4,8,17,18]. BD Microtainer® products are approved for in vitro diagnostic use and showed great acceptability among donors who collected an average of 0.4 mL of capillary blood (10–15 drops) in less than 5 min. However, lack of standardization prevents consumers from using microtubes without laboratory validation of preanalytical conditions. Our study included 12 routine analytes, but excluded electrolytes, bicarbonate, bilirubin, and albumin. Glucose was incompatible with analysis in uncentrifuged microsamples but replaced for Hemoglobin A1c performed in whole blood. All other assays were performed in capillary plasma, and variability was expected as capillary blood is a mixture venous blood, arterial blood, and tissue fluid. Additionally, venous serum values have been reported to slightly differ from those obtained from capillary serum [19]. As shown in Fig. 2, this study demonstrates interchangeability in most chemistry values obtained from capillary plasma collected in lithium-heparin and equivalent to results obtained from microsamples collected in an alternative anticoagulant [20]. Excluding AST, small differences observed are not important during health screenings where laboratory tests are only utilized to identify a subset of the population who should have an additional evaluation for adequate medical diagnosis. We demonstrated satisfactory results for most analytes despite mechanical stress, high temperature and long delays in centrifugation. There was no hemolytic interference as opposed to previous reports using traditional venous samples [21]. Lack of agreement was observed for AST due to its lower allowable level of hemoglobin index in uncentrifuged microsamples [22]. Total protein shows poor correlation due to the short dynamic range of the study cohort, but microsample assay results have an excellent agreement against reference values in each case.

5. Conclusions

This study provides a proof-of-concept about using Microtainers® for capillary blood collection and analysis under uncontrolled preanalytical variables such as temperature and delayed centrifugation of samples transported over a 24 h timeline. This study may be useful to define acceptable delay times and storage conditions for other laboratories when a short time between sample collection and processing is not possible in large-scale health screening events or at-home blood sample collection.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CRediT authorship contribution statement

Cristian Saez: Conceptualization, Investigation, Formal analysis, Visualization, Writing - original draft. Elizabeth Fontaine: Investigation, Writing - review and editing.

Declaration of competing interest

The authors do not have any conflict to report. There are no commercial interests between the authors and BD Diagnostics.

Data availability

Data will be made available on request.