Home Bias, its minimization or circumvention to simplify internal quality assurance
Article Open Access

Bias, its minimization or circumvention to simplify internal quality assurance

An erratum for this article can be found here: https://doi.org/10.1515/labmed-2017-0132
  • Rainer Haeckel EMAIL logo , Eberhard Gurr , Torsten Hoff and on behalf of the working group Guide Limits of the German Society of Clinical Chemistry and Laboratory Medicine (DGKL)
Published/Copyright: July 30, 2016
Download Article (PDF)
Become an author with De Gruyter Brill
Submit Manuscript Author Information
LaboratoriumsMedizin
From the journal LaboratoriumsMedizin Volume 40 Issue 4

Abstract

Several concepts of analytical bias and remedies to minimize bias have been suggested with the ultimate goal to disregard it. Short-term bias (within one control cycle) should be treated as a random error if it is less than the permissible limits. Long-term bias should be eliminated if it is known or circumvented by estimating intra-laboratory reference limits (RLs). Consequently, analytical uncertainty could be reduced to permissible imprecision. Then, models combining imprecision and bias would become irrelevant, and the numerical value of total analytical error would become identical with imprecision. The purpose of the present report is to simplify quality assurance schemes considerably by disregarding bias either by estimating RLs or by verifying the applied reference limits (checking the transferability) as requested by ISO and CLSI.

Keywords: bias; internal quality assurance; measurement uncertainty

Abbreviations: B, bias; pB, permissible bias; GUM, Guide to measurement uncertainty; MU, measurement uncertainty; CVA, analytical coefficient of variation; CVB, biological coefficient of variation; CVC, combined intra- and inter-individual coefficient of variation; CVE, empirical CVB; CVE*, empirical CVB derived from the logarithmic scale; CVG, inter-individual coefficient of variation; CVI, intra-individual coefficient of variation; sA, analytical standard deviation; EL, equivalence limit; EQAS, external quality assessment scheme; psA, permissible analytical standard deviation; psA,Xi, permissible analytical standard deviation of a particular measured value; pCVA, permissible analytical CVA; PL, permissible limit; RI, reference interval; RL, reference limit; RL1, lower reference limit (RL2.5); RL2, upper reference limit (RL97.5); RiliBÄK, Richtlinie der Bundesärztekammer, official German guidelines; RMSD, root mean square of measurement deviation; TAE, total analytical error; U, expanded uncertainty; pU, permissible expanded uncertainty; VIM, vocabulaire international de métrologie; WG, working group Guide Limits (Richtwerte) of the German Society of Clinical Chemistry and Laboratory Medicine (DGKL).

Classical concept of analytical error in laboratory medicine

Minimizing analytical errors in the interest of patient safety is a prerequisite for a well-functioning medical laboratory [1]. The classical concept defined several decades ago [2], is still valid and useful for quality assurance of the analytical phase. The classical concept classifies two major components: random and systematic error. In the more recent concept of measurement uncertainty, the dispersion of a set of quantity values for a measurand (the quantity intended to be measured) is also attributed to random and systematic effects [3].

Random error

Random error is

  1. the dispersion of analytical results around a measured value (imprecision),

  2. determined as standard deviation from a defined number of replicate measurements of the same measurand carried out under specified conditions [3] and may be expressed in percent of the mean value (relative standard deviation),

  3. unavoidable,

  4. accepted or should be accepted by requesters of laboratory examinations (e.g. physicians) and is expected or should be expected by requesters to be kept as small as possible.

The imprecision is the “numerical expression of precision” [4]. According to Guide to Measurement Uncertainty (GUM) [5], imprecision is termed standard uncertainty of a single measurement (us):

uS=sA

In the RiliBÄK [1] guideline for quality assurance of laboratory medical examinations, the term empirical standard deviation is used [6]. The standard uncertainty is usually determined with control materials (with a matrix comparable to human samples as much as possible) as standard deviation sA from day to day (intra-laboratory reproducibility, replicate measurements under intermediate precision conditions). It is common practice to choose a 1-month period (about 20 working days) [7].

Systematic error

Systematic errors are a heterogeneous group of errors which occur in one direction (uni-directional) in contrast to random errors which varies in two directions around a mean value (bi-directional). Systematic errors cannot be explained by one single algorithm. Several types of systematic errors are described in the literature [8], [9] and are summarized in Table 1. The most common systematic error in quality assurance schemes is bias (B).

Table 1:

Various intra-laboratory measurement types of systematic errors.

Type of systematic errorPreferred sampleConsidered by RiliBÄKConsidered by WGa conceptExamples
1. Bias
Shift (“Sudden jumps”)CMYesYesCalibration, lot-to-lot variation of reagents
Permanent biasCM, PSYesYesInappropriate calibrator, heterogenity of antibodies
Drift
 Short-term driftCMYesbYesbReagent instability (including calibrators)
 Long-term driftCM, PSNoYesReagent instability (including calibrators)
2. Carry-overPSNoNo
3. Non-specificitiesCMYesbYesbPseudo-creatinines
PSNoNoPseudo-creatinines
Selected PSNoNoInterferences (e.g. drugs, lipemia, etc.)

aWG, working group Guide Limits. bIf control limits are surpassed. Preferred samples for detecting the various error types are either patients samples (PS) and/or control materials (CM).

Bias is

  1. defined as the difference of the mean that would ensue from an undefined number of replicated measurements of the same measurand carried out under specified conditions and the true value of the measurand [4], bias may also be expressed as a fraction of the true value (relative bias) or as %bias in percent of the true value,

  2. often avoidable, respectively somehow correctable,

  3. usually not expected by requesters.

In laboratory medicine, true values generally are unknown. Target values for control materials are applied instead of true values. Target values are preferably determined by a reference method (reference method values) or by the manufacturer with a standardized procedure (method assigned values). If a control material has reference method values, they should be taken as target values, otherwise assigned values must be used.

Furthermore, bias can be determined using comparison of measurement results with results achieved with a reference method or a well-accepted standard method (method comparison study), and by inter-laboratory comparisons (external quality assessment schemes – EQAS-, proficiency testing, round robin testing). In the case of a comparison study with samples of different quantities, the difference between methods indicates a bias difference of the individual methods. If the comparative method is not a reference method, the difference between the two methods is not considered as the bias of the test method. It is the difference between the two methods and is identical with the average of the differences. In the case of a regression equation, slope and intercept represent proportional and constant error components of bias.

Physicians usually accept that laboratory data vary and have unavoidable random errors, but they are not acquainted in considering bias. Analytic bias can have profound effects near clinical decision points [9], [10], [11].

Bias may appear as a sudden shift or as an increasing or decreasing trend (drift). A shift may occur suddenly or is permanently present. As an example, a calibrator or a reagent may deteriorate increasingly with time (causing a trend/drift of the results) or suddenly (causing a shift/“jump”). Lot-to-lot variation may be treated as random error (included in imprecision), as long as it occurs within the permissible imprecision. Otherwise, it must be eliminated by corrective actions (e.g. by recalibration). However, any trimming of the dataset should be carefully justified [4]. Causes of a permanent bias may be poorly defined or unstable calibrators (reference calibrators not available) or reagents (e.g. antibodies).

A drift may occur during one control cycle (short-term drift) or may only be detectable during a longer time period (long-term drift). The observation period should be related to the period between reviews of reference limits or the time period using a control material batch. A sudden bias or a drift can be easily detected on a Shewhart-type control chart. In the former RiliBÄK [12], a continuous increase or decrease of seven subsequent measurement values was considered as a significant indicator of a trend.

Long-term drifts could influence reference limits. Therefore, periodic determination of reference limits can play an important role in a quality assurance system. Long-term analytical system instability is often neglected. Long-term drifts can be observed with the same lot of control materials or with samples of patients. Means of control cycles are presented graphically and visually inspected for trends or shifts. A simple test is suggested below. A long-term trend has been observed with patients samples over years for glucose measurements with a 5 years’ increase of 0.55 mmol/L (11%) [13]. Long-term instability may lead to serious problems which often are neglected in practice. It can cause erroneous medical decisions if reference limits are not adjusted accordingly [9], [14]. Such trends are usually not included in measurement uncertainty (MU) concepts. However, these bias effects may be considered if reference limits are periodically estimated or reviewed by an indirect procedure [15], [16]. Long-term drift effects can also be detected from external quality assessment schemes [17]. If suspected, this bias type must be investigated and be eliminated.

The consideration of bias is only required for comparisons between laboratories (comparability) and if reference limits are taken from external sources. Comparability between laboratories may be sufficiently assured by regular participation in EQAS and by using CE-marked analytical procedures. For the majority of their measurements, most routine laboratories in Europe usually apply standardized analytical procedures which follow the strict requirements of the EC directive 98/78 EU. “Future developments in reference measurement systems are likely to continue to play a major role in minimizing bias in clinical chemistry. Reference measurement systems are, however, unlikely to solve all the most complex bias issues, for example, in the fields of immunochemistry” [1]. Total comparability between laboratories probably remains an elusive goal as was recently pointed out [18], [19].

The goal should be to keep bias as low as possible. Two ways are possible to disregard bias: 1. A high degree of comparability by continuous standardization makes surveillance of bias less relevant or even superfluous. 2. Using intra-laboratory reference limits (RLs) to circumvent the numerous problems with transferability [16], [19]. Single laboratories may establish their own reference limits assuming that they are constant, or remain stable until they are reviewed. Adjusting reference limits should be preferred instead of correcting measured values due to an established bias. Many approaches have been developed to derive intra-laboratory RLs from the large data pools usually stored in laboratory information systems. One of the latest algorithms was published by Arzideh et al. [15], [16]. A software program accessible with a Microsoft Excel spread sheet is available from the home page of the German Society of Clinical Chemistry and Laboratory Medicine (DGKL) or from the authors. If intra-laboratory RLs are established, only random components of permissible expanded uncertainty (pU) are to be considered and internal quality assurance could be simplified considerably.

Non-specificities of measurement procedures caused by sample specific interferences have been called “random error” [8], although their character is systematic. They are not considered as bias in quality assurance schemes. If they are quasi-constant, they may be treated as a bias. A typical example are pseudo-creatinines (a mixture of interferents). A fixed amount of 27 μmol/L has been proposed for adults to correct creatinine results obtained by an unspecific Jaffé reaction [16]. Such quasi-constant non-specificities should be identified and considered (e.g. using a fixed amount or by specific reference limits).

Permissible limits of bias

Whereas short-term bias may be included in the estimation of imprecision to some extent, long-term trends should be either eliminated or treated separately. In a recent proposal for permissible uncertainty [20], [21], the permissible bias (pB) was referred to the permissible imprecision (psA) and set to 0.5·psA, because B=0.5·sA leads to about the same percent of false-positive results (Figure 1) as an imprecision of one sA [19].

Figure 1: The effect of bias and imprecision on Δ false-positive results (ΔFPR) for the example of a biological variation CVB=20%.Crosses: CVA (1%–10%, bias=0), triangles: bias (1%–10%, CVA=0), rhombs: bias+CVA=10% (relation between bias and imprecision altered in 0.1 steps starting with CVA=10%/bias=0 up to CVA= 0/bias=10%), rectangles: theoretical case that bias and imprecision would affect ΔFPR equally (CVA+bias=10%). Taken from Ref. [19].
Figure 1:

The effect of bias and imprecision on Δ false-positive results (ΔFPR) for the example of a biological variation CVB=20%.

Crosses: CVA (1%–10%, bias=0), triangles: bias (1%–10%, CVA=0), rhombs: bias+CVA=10% (relation between bias and imprecision altered in 0.1 steps starting with CVA=10%/bias=0 up to CVA= 0/bias=10%), rectangles: theoretical case that bias and imprecision would affect ΔFPR equally (CVA+bias=10%). Taken from Ref. [19].

According to the GUM concept [5], the uncertainty of the bias estimation should be considered. Recently [20], we had proposed to set this uncertainty to uB=0.5·psA and to expand the pB to

(1)pB=[(0.5·psA)2+(0.5·psA)2]0.5=0.7·psA

pB%=0.7⋅pCVA

Permissible standard uncertainty (psA or pCVA) can be either established as state-of the-art (concept 2, see below) or may be derived from biological variation (CVB) [20]. As a surrogate for CVB, we recommended the empirical combined biological variation CVE* and proposed a non-linear relation between CVE* and pCVA [20]:

pCVA= (CVE∗–0.25)0.5

psA=pCVA⋅median⋅0.01

In the proposed approach [22], psA and pCVA are derived from pCVA at the median of the reference interval for each measured value xi (sA,Xi and CVA,Xi) and equation 1 becomes pB=0.7·psA,Xi. Other concepts (as e.g. TAE, RiliBÄK) assume a constant CVA within the measurement interval. Recently, we had pointed out that CVA is never constant in the measurement interval [20].

As mentioned above, long-term bias can be observed with the same lot of control materials. Means of control cycles are presented graphically and visually inspected for trends or shifts. A simple test for clinical relevance of an observed bias is to establish equivalence limits [23]. Assuming that the variation of the mean values over a long term is not known a priori and is constant, the equivalence limits may be based on the pB. According to equation 1, the proposed equivalence limits (EL) are

(2)EL=mean1±1.96⋅0.7⋅psA=mean1±1.37⋅psA,Xi

In this equation, mean1 is the mean value of the first control cycle if a new lot of the control material is started, psA,Xi is the permissible analytical standard deviation corresponding to xi=mean1. If the regression line through the mean values exceeds the lower or upper equivalence limit, the bias may become relevant and should not be tolerated anymore.

Long-term bias can also be observed with measurement values of patients. It has been shown that glucose measurements became increasingly biased upwards as time passed (see above). If monthly median values of patients are graphically presented, they can be interpreted in analogy to control samples by equation 2. This approach may not be suited for measurands with large reference intervals or if only one reference limit is available.

Problems with combining imprecision and bias in quality assurance

Several concepts for combining random and systematic errors for the purpose of quality assurance and their permissible limits (PL), often termed control limits, are often controversially discussed:

  1. Total analytical error (TAE) [24]

    PL=target value±(z⋅psA+|pB|)

    The statistical coverage factor z chosen may be=1.64 [25] or 1.96 [24]. psA and pB are usually based on the biological variation (biological variation model).

  2. RiliBÄK [6]:

    1. Measurands listed in Table B 1a–c of the RiliBÄK [6]

      PL=target value±RMSD

    2. Measurands not listed

      PL=target value±Δmax

      Δmax=(z2⋅ sep2+δep2)0.5

    RMSD means root mean square of measurement deviation (column 3 in Table B 1a–c of the RiliBÄK) [6]. The limits have been established based of surveys with invited laboratories in order to cover 90 or 95% of all participants (state-of-the-art model) [20]. Δmax is the maximum permissible error when measuring a control sample, sep is the empirical standard deviation, z is 3 and δep (bias) is the mean concentration measured in the control sample minus the target value of the control sample provided by the manufacturer.

  3. MU approach [7], [20]:

    PL=target value±pU

    (3)pU=z⋅(psA,Xi2+pB2)0.5

According to equation 1, equation 3 can be rearranged to

(4)pU=z⋅(psA,Xi2+0.72⋅psA,Xi2)0.5= 1.96⋅1.22⋅psA,Xi=2.39⋅psA,Xi

Bias should not be included in the uncertainty estimation if B>0.7·sA and must be eliminated by corrective actions (e.g. by recalibration).

In all three concepts, the target values provided by the manufacturer of the control material are used. The TAE concept adds the two components linearly. The RiliBÄK concept adds the components as variances as in the theorem of Pythagoras [26].

All concepts have their merits and disadvantages. Concept 3 is a compromise between model 1 which is based on biological variation and concept 2 which is primarily a state-of-the-art approach as outlined recently [20].

If imprecision and bias are combined to define permissible limits, both components can compensate each other, that means a very low (minimized) bias allows a larger pCVA (and vice versa). Then, the pCVA may become too large for diagnostic purposes. Therefore, it is necessary that pCVA and pB are assured and limited independently [19], [26], [27] as, e.g. recommended in the former RiliBÄK and in our recent proposal [7], [20]. In Vocabulaire International de Métrologie (VIM), a combined error model is not mentioned: “random and systematic errors have to be treated independently” [3]. “If bias for a measurement method or system is known, it is difficult to see the logic in including it in the calculation of the total error rather than eliminating it by re-calibration. If the bias cannot be determined, it is unknown and cannot be eliminated”[1].

Quality assurance disregarding bias

Imprecision and bias have different and complex effects on the diagnostic sensitivity (false positive rates) as demonstrated in Figure 1. The rate of false-positive results is more highly dependent on bias than on imprecision. Therefore, both error types should not receive equal weights, as e.g. in the latest RiliBÄK. A correct algorithm for combining imprecision with the various bias types would be very complicated. Therefore, the ideal solution is to eliminate bias, either by improving the methodology to a degree that bias becomes negligibly small, or by correcting bias or by circumventing bias. Then, numerically pTAE=RMSD=pU=z·pCVA, that means only pCVA is needed for quality assurance schemes.

Concept 3 can be simplified if bias can be disregarded (as already mentioned above):

(5) PL=mean1±z⋅psA,Xi

In equation 5, mean1 is the average of replicate measurements during the first control cycle of a control period and z=1.96. A control period usually is defined by the “life-time” of a control material batch. Target values are not required for internal quality assurance in the simplified concept 3, however, they remain important for external quality assurance and for estimating the amount of bias if it is detected. For the first control cycle (RiliBÄK: “Ermittlungsperiode”) of a control material batch, target value and pCVA recommended by the manufacturer can be applied. After the first control cycle (about 20 days), the mean of the replicate control sample values from the first cycle is calculated. A control chart is constructed with the mean and the control limits (1.96·psA,xi). Control charts of the following control cycles always use the mean of the previous cycle. If a new lot of the control material is started, it is common practice to overlap the first control cycle with the last cycle of the previous lot.

The transferability of RLs should define the requirement of considering bias as summarized in the Table 2. Approach 1 and 2 are favored by the authors. If RLs are not verified, approach 3c and especially 3d may be applied. Approach 3a and 3b are required by the present RiliBÄK.

Table 2:

Permissible limits of quality assurance depending on the transferability of reference limits, comparing the approaches of RiliBÄK and of the working group Guide Limits of the German Society for Clinical Chemistry and Laboratory Medicine (WG).

Type of reference limits (RLs)TransferabilityControl limitsaBiasSource
1. Intra-laboratory RLsNo problemsMeanb±1.96·pCVATo be neglectedWG
2. Extra-laboratory RLsVerifiedMeanb±1.96·pCVA To be neglectedWG
3. Extra-laboratory RLsNot verifieda) Target valuec±RMSD (I and B combinedd)To be assumedRiliBÄK
b) Target valuec±[(9·sep2+δepb)0.5]To be assumedRiliBÄK
c) Target valuec±pU% (I and B combined)To be assumedWG

aThe ±limits given as coefficient of variation must be converted to the corresponding standard deviation. bMean value of replicates measured during the previous control cycle. cTarget value provided by the manufacturer of the control material (reference method value or assigned value). dI, imprecision and B, bias. eImprecision assured as in approach 1, bias assured after each control cycle (target value – mean value).

Discussion

Long-term stability of measurement results may become a complex problem if reagent lots, control material lots, pre-examination protocols, and especially if analytical procedures are changed. Although such causes for bias are excluded by the present concept, they may be tolerated to some extent if they occur within the pB limits. Long-term bias should be observed either by control materials as suggested above or by our indirect approach for estimating reference limits including checking for long-term stability as part of a patient-based quality assurance scheme. It has been shown, that long-term stability of analytical systems can be supervised more reliably by patients’ samples than by commercial quality control materials [28]. Observing patient percentiles has also been suggested as a useful quality management tool for selected measurands by many authors [29], [30], [31].

If the laboratory applies different analytical procedures with different bias for the same measurand, it should use different reference limits corresponding to the analytical procedure. Differences between laboratories can be disregarded if the same analytical procedure (including instruments, calibrators and reagents) are applied and the laboratories have shown that they can use common reference limits [32]. Relevant differences between laboratories must be assumed if common reference limits have not been established.

Measurement procedures are validated and should be introduced in routine service by laboratories not before verification. Reference limits have been either established or at least verified by the laboratories. This chain of consecutive steps is usually considered as a static system (“frozen” forever [33]). However, this assumption may not be true during the lifecycle of many procedures. Probably, methodology is often prone to a non-static, evolutionary process, and reference limits determined decades ago may not be appropriate for ever. Therefore, it appears essential to review reference limits regularly as claimed by ISO 15189 [34]. Such a reviewing and eventual adaptation of the reference limits would take possible dynamics of measurement procedures into account during their lifetime.

In contrast to random error, systematic errors are a complicated mixture of different effects which are difficult to classify and which cannot be correctly treated with one single algorithm. Therefore, it appears useful to get rid of the bias problems somehow, e.g. by establishing RLs. We have postulated that laboratory information systems should include algorithms to establish reference limits from stored patients’ measurement values [20].

It is interesting that all concepts for combining analytical errors become very similar if bias is expressed as a fraction of imprecision. Fraser changed his original equation for bias [35] to B=0.25·(CVI2+CVG2)0.5 [25]. Assuming the optimal case [35], that the intra-individual CVI is close to the inter-individual CVG, and that CVA=0.5·CVI, TAE becomes 1.64·0.5·CVI+0.25·(CVI2+CVI2)0.5=2.35·pCVA which is similar to pU%=2.39·pCVA (combination concept 3). With combination concepts 3a and 3c (Table 2), the average permissible limits are also very similar.

A limitation of the present proposal is that possible non-commutablities of control materials, as e.g. “matrix-related bias” [30] are not considered. These effects may become relevant for procedures with a small imprecision. Methods with poor precision can be expected to be less sensitive to these effects [31]. Another, not considered problem is that lot-to-lot variation may affect the results of human samples but is not detected by quality assurance schemes with control materials [36].

Conclusions

Several types of bias occur, and there is no way to correctly consider all types with one algorithm. The visionary goal should be to reduce bias to a negligible level by improving methodology. However, this goal is presently rather idealistic. Short-term bias may be unavoidable and can be included in imprecision if the resulting sum is less than the permissible analytical imprecision. Long-term bias should be either eliminated or circumvented, e.g. by RLs. If a laboratory uses different analytical procedures, reference limits must be established for each procedure.

Limits for random error should be derived of biological variation considering that pCVA is not constant in the measuring interval as realized by the uncertainty approach of the DGKL working group Guide Limits [7], [20].

  1. Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.

  2. Research funding: None declared.

  3. Employment or leadership: None declared.

  4. Honorarium: None declared.

  5. Competing interests: The funding organization(s) played no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.

References

1. Theodorsson E, Magnusson B, Leito I. Bias in clinical chemistry. Bioanalysis 2014;6:2855–75.10.4155/bio.14.249Search in Google Scholar PubMed

2. Büttner, Borth R, Boutwell JH, Broughton PM, Bowyer RC. Approved recommendation (1978) on quality control in clinical chemistry. J Clin Chem Clin Biochem 1980;18:69–77.Search in Google Scholar

3. International vocabulary of metrology (vim) – basic and general concepts and associated terms. Third edition. Jcgm 200; 2012, http://www.bipm.org/vim.Search in Google Scholar

4. Clinical and Laboratory Standards Institute. Expression of measurement uncertainty in laboratory medicine; approved guideline. CLSI document C51-A. Wayne, PA: CLSI, 2012;vol.32, No.4.Search in Google Scholar

5. International Organisation for Standardisation. Guide to the expression of uncertainty in measurement. Genf, ISBN 92-67-10188-9, first edition 1993, corrected and reprinted 1995.Search in Google Scholar

6. Richtlinie der Bundesärztekammer zur Qualitätssicherung quantitativer laboratoriumsmedizinischer Untersuchungen. Dt Aerzteblatt 2014;111:A1583–1618.Search in Google Scholar

7. Haeckel R, Gurr E, Wosniok W, Peil B. Supplements to a recent proposal for permissible uncertainty of measurements in laboratory medicine. J Lab Med 2016;40:141–5.10.1515/labmed-2015-0112Search in Google Scholar

8. Krouwer JS. Setting performance goals and evaluating total analytical error for diagnostic assays. Clin Chem 2002;48:919–27.10.1093/clinchem/48.6.919Search in Google Scholar

9. Magnusson B, Ellison SL. Treatment of uncorrected measurement bias in uncertainty estimation for chemical measurements. Anal Bioanal Chem 2008;390:201–13.10.1007/s00216-007-1693-1Search in Google Scholar PubMed

10. Clinical Laboratory Standard Institute C24A3. Statistical quality control for quantitative measurement procedures: principles and definitions. CLSI, Wayne, PA, 2006.Search in Google Scholar

11. Klee GG. Establishment of outcome-related analytic performance goals. Clin Chem 2010;56:714–22.10.1373/clinchem.2009.133660Search in Google Scholar PubMed

12. Richtlinie der Bundesaerztekammer zur Qualitätssicherung laboratoriumsmedizinischer Untersuchungen. Dt Aerzteblatt 2008;105:C301–13.Search in Google Scholar

13. Froslie KF, Godang K, Bollerslev J, Henriksen T, Roislien J, Veierod MB, et al. Correction of unexpected increasing trend in glucose measurements during 7 years recruitment to a cohort study. Clin Biochem 2011;44:1483–6.10.1016/j.clinbiochem.2011.08.1150Search in Google Scholar PubMed

14. Coucke W, van Blerk M, Libeer JC, van Campenhout C, Albert A. A new statistical method for evaluating long-term analytical performance of laboratories applied to an external quality assessment scheme for flow cytometry. Clin Chem Lab Med 2010;48:645–50.10.1515/CCLM.2010.122Search in Google Scholar PubMed

15. Arzideh F, Wosniok W, Gurr E, Hinsch W, Schumann G, Weinstock N, et al. A plea for intra-laboratory decision limits. Part 2. A bimodal deductive concept for determining decision limits from intra-laboratory data bases demonstrated by catalytic activity concentrations of enzymes. Clin Chem Lab Med 2007;45:1043–57.10.1515/CCLM.2007.250Search in Google Scholar PubMed

16. Arzideh F, Wosniok W, Haeckel R. Reference limits of plasma and serum creatinine concentrations from intra-laboratory data bases of several German and Italian medical centres. Comparison between direct and indirect procedures. Clin Chem Acta 411;2010:215–21.10.1016/j.cca.2009.11.006Search in Google Scholar PubMed

17. Matar G, Poggi B, Meley R, Bon C, Chardon L, Chikh K, et al. Uncertainty in measurement for 43 biochemistry, immunoassay, and hemostasis routine analytes evaluated by a method using only external quality assessment data. Clin Chem Lab Med 2015;53:1725–36.10.1515/cclm-2014-0942Search in Google Scholar PubMed

18. Boyd JC. Cautions in the adoption of common reference intervals. Clin Chem 2008;54:238–9.10.1373/clinchem.2007.098228Search in Google Scholar PubMed

19. Haeckel R, Wosniok W. A new concept to derive permissible limits for analytical imprecision and bias considering diagnostic requirements and technical state-of-the-art. Clin Chem Lab Med 2011;49:623–35.10.1515/CCLM.2011.116Search in Google Scholar PubMed

20. Haeckel R, Gurr E, Wosniok W, Peil B. Permissible limits for uncertainty of measurement in laboratory medicine. Clin Chem Lab Med 2015;53:1161–71.10.1515/cclm-2014-0874Search in Google Scholar PubMed

21. Haeckel R, Wosniok W, Streichert T. Optimizing the use of the “state-of-the-art” performance criteria. Clin Chem Med Lab. 2015;53:887–91.10.1515/cclm-2014-1201Search in Google Scholar PubMed

22. Haeckel R, Wosniok W, Arzideh F. Equivalence limits of reference intervals for partitioning of population data. Relevant differences of reference limits. J Lab Med 2016;40:199–205.10.1515/labmed-2016-0002Search in Google Scholar

23. Haeckel R, Gurr E, Keller T. Permissible measurement uncertainty in the lower part of measurement intervals. J Lab Med 2016;40:277–82.10.1515/labmed-2016-0006Search in Google Scholar

24. Westgard JO. Update on measurement uncertainty: new CLSI C51A guidance. www.westgard.com/clsi-c51.htm, assigned 2/24/2012.Search in Google Scholar

25. Fraser CG, Petersen PH. Quality goals in external quality assessment are best based oin biology. Scand J Clin Lab Invest 1993;53(suppl. 212):8–9.10.1080/00365519309085446Search in Google Scholar

26. Haeckel R, Wosniok W. Benefits of combining bias and imprecision in quality assurance of clinical chemistry procedures. J Lab Med 2007;31:89–9.10.1515/JLM.2007.016Search in Google Scholar

27. Hylthof Petersen P, Klee P. Influence of analytical bias and imprecision on the number of false positive results using Guideline-Driven Medical Decision Limits. Clin Chim Acta 2014;430:1–8.10.1016/j.cca.2013.12.014Search in Google Scholar PubMed

28. Feinberg M, Boulanger W, Dewe W, Hubert Ph. New advances in method validation and measurement uncertainty aimed at improving the quality of chemical data. Anal Bioanal Chem 2004;380:502–14.10.1007/s00216-004-2791-ySearch in Google Scholar PubMed

29. Van Houcke S, Stepman HC, Thienpont LM, Fiers T, Stove V, Couck P, et al. Long-term stability of laboratory tests and practical implications for quality management. Clin Chem Lab Med 2013;51:1227–31.10.1515/cclm-2012-0820Search in Google Scholar PubMed

30. Wilson A, Roberts WA, Pavlov I, Fontenot J, Jackson B. Patient result median monitoring for clinical laboratory quality control. Clin Chim Acta 2011;412:1441–9.10.1016/j.cca.2011.04.024Search in Google Scholar PubMed

31. Schimnich AA, Bruns DE, Boyd JC, Bryant SC, La Fortune KA, Grebe SK. Failute to current laboratory protocols to detect lot-to-lot reagent differences: findings and possible solutions. Clin Chem 2013;59:1187–94.10.1373/clinchem.2013.205070Search in Google Scholar PubMed

32. Zierk J, Arzideh F, Haeckel R, Cario H, Frühwald MC, Groß HJ, et al. Common pediatric reference intervals for alkaline phosphatase. Clin Chem Lab Med DOI: 10.1515/cclm-2016-0318. [Epub ahead of print].Search in Google Scholar PubMed

33. Miller WG, Erek A, Cunningham TD, Oladipo O, Scott MG, Johnson RE. Commutability limitations influence quality control results with different reagent lots. Clin Chem 2011;57:76–83.10.1373/clinchem.2010.148106Search in Google Scholar PubMed

34. International Standard Medical Laboratories – Particular requirements for quality and competence, ISO 15189-2003(E),1–39.Search in Google Scholar

35. Fraser CG. Biological variation: from principles to practice. Washington DC, AACC Press, 2001:1–151.Search in Google Scholar

36. Clinical and Laboratory Standards Institute. User evaluation of between-reagent lot variation; approved guideline. CLSI document EP26-A. Wayne, PA: CLSI, 2013;vol.33, No.12.Search in Google Scholar

Received: 2016-5-25
Accepted: 2016-7-6
Published Online: 2016-7-30
Published in Print: 2016-8-1

©2016 Walter de Gruyter GmbH, Berlin/Boston

This article is distributed under the terms of the Creative Commons Attribution Non-Commercial License, which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Articles in the same Issue

  1. Frontmatter
  2. Editorial
  3. Wechsel in der Schriftleitung: Peter Schuff-Werner übernimmt die Schriftleitung bei der Zeitschrift LaboratoriumsMedizin
  4. Labormanagement/Laboratory Management / Redaktion: E. Wieland
  5. Chancen und Risiken von e-Health in der Labormedizin
  6. Infektiologie und Mikrobiologie (Schwerpunkt Bakteriologie)/ Infectiology and Microbiology (Focus Bacteriology) / Redaktion: P. Ahmad-Nejad/B. Ghebremedhin
  7. Normal and abnormal vaginal microbiota
  8. Pädiatrisches Labor/Pediatric Laboratory Medicine / Redaktion: K.P. Kohse
  9. Association of osteocalcin, insulin resistance and oxidative stress during noncomplicated pregnancy
  10. Originalarbeiten/Original Articles
  11. Flow cytometry as an important tool in the diagnosis of immunodeficiencies demonstrated in a patient with ataxia-telangiectasia
  12. Bias, its minimization or circumvention to simplify internal quality assurance
  13. Permissible measurement uncertainty in the lower part of measurement intervals
  14. Efficient use of laboratory resources: pre-screening for urine cultures by automated urinalysis and microscopy to allow exclusion of specimens from culture workflow
  15. Comparison of student perception and performance between case-based learning and lecture-based learning in a clinical laboratory immunology course
https://doi.org/10.1515/labmed-2016-0036
Keywords for this article
bias; internal quality assurance; measurement uncertainty
Creative Commons
BY-NC-ND 3.0

Articles in the same Issue

  1. Frontmatter
  2. Editorial
  3. Wechsel in der Schriftleitung: Peter Schuff-Werner übernimmt die Schriftleitung bei der Zeitschrift LaboratoriumsMedizin
  4. Labormanagement/Laboratory Management / Redaktion: E. Wieland
  5. Chancen und Risiken von e-Health in der Labormedizin
  6. Infektiologie und Mikrobiologie (Schwerpunkt Bakteriologie)/ Infectiology and Microbiology (Focus Bacteriology) / Redaktion: P. Ahmad-Nejad/B. Ghebremedhin
  7. Normal and abnormal vaginal microbiota
  8. Pädiatrisches Labor/Pediatric Laboratory Medicine / Redaktion: K.P. Kohse
  9. Association of osteocalcin, insulin resistance and oxidative stress during noncomplicated pregnancy
  10. Originalarbeiten/Original Articles
  11. Flow cytometry as an important tool in the diagnosis of immunodeficiencies demonstrated in a patient with ataxia-telangiectasia
  12. Bias, its minimization or circumvention to simplify internal quality assurance
  13. Permissible measurement uncertainty in the lower part of measurement intervals
  14. Efficient use of laboratory resources: pre-screening for urine cultures by automated urinalysis and microscopy to allow exclusion of specimens from culture workflow
  15. Comparison of student perception and performance between case-based learning and lecture-based learning in a clinical laboratory immunology course
Sign up now to receive a 20% welcome discount
Institutional Access
How does access work?
Have an idea on how to improve our website?
Please write us.
© 2025 De Gruyter Brill
Downloaded on 31.10.2025 from https://www.degruyterbrill.com/document/doi/10.1515/labmed-2016-0036/html?lang=en
Scroll to top button