Quality specifications in EQA schemes: from theory to practice

Clinica Chimica Acta 346 (2004) 87 – 97 www.elsevier.com/locate/clinchim

Quality specifications in EQA schemes: from theory to practice Laura Sciacovelli *, Lorena Zardo, Sandra Secchiero, Mario Plebani Centro di Ricerca Biomedica, Via Ospedale 18, 31033 Castelfranco Veneto (TV), Italy Received 13 February 2004; accepted 22 February 2004

Abstract Background: External quality assessment (EQA) is a tool for quality management in clinical laboratories and its main objectives are assessment of participants and methods performance, training and advice. This paper describes the quality specifications used in EQA schemes of the Centre of Biomedical Research (CRB), in order to design schemes that can assess laboratory reliability performances, meet the changing needs and quality recommendations. Methods: Quality specifications for control materials, statistical procedures and goals to assess laboratory performance have been applied and introduced in EQA schemes managed by CRB. Results: The application of well-defined quality specifications has demonstrated effective. In particular, we report results on alkaline phosphatase and cholesterol obtained using commercial control materials and human serum controls, in two different EQA surveys; the inter-laboratory variability (CVinter%) for troponin I analysed with a diagnostic system and assigned values of CK-MB mass obtained using four different diagnostic systems; the percentage of acceptable performances obtained by means of the application of goals based on clinical criteria, biological variation, state-ofthe-art and used for EQA schemes, and referring to some analytes with significant clinical values such as cholesterol, glucose, glycated hemoglobin and sodium. Conclusions: The design of reliable EQA schemes based on evidence-based quality specifications is a pre-requisite for supporting the quality improvement of clinical laboratories. D 2004 Elsevier B.V. All rights reserved. Keywords: External quality assessment; Analytical performance laboratory; Analytical total error; EQA control material; EQA statistical procedure

1. Introduction Medical laboratories must provide high quality service to clinicians by producing accurate, precise, relevant and comprehensive data that can be applied to the medical management of patients. They must, therefore, produce analytically reliable results, and the information required for the correct interpretation and

* Corresponding author. Tel.: +39-423-732823; fax: +39-423732826. E-mail address: [email protected] (L. Sciacovelli). 0009-8981/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.cccn.2004.02.037

use of results. The mechanism for achieving these objectives is encompassed in total quality management [1– 5]. The analytical process and its control are becoming increasingly reliable thanks to the improvements made to instruments and methods, and the application of internal quality control (IQC) and external quality assessment (EQA) procedures. However, the large variety of laboratory methods available can hamper measurements and the comparability of results, thus potentially compromising patient management. Medical laboratories have a long tradition of EQA procedures but, continuous progress made in labora-

88

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

tory medicine, imposes a constant development and change in the EQA design. As a logical evolution in quality management, EQA organizers must arrange their schemes according to quality specifications in order to constantly encourage improvement in laboratory to higher and higher standards, thus striving to meet laboratories’ needs. In the modern medical laboratory, the main objectives in using the EQA programs are: to assess participants’ performance and method performance, and provide training and advice. The EQA, which should be considered a tool for quality management within the ambit of total quality management, must help laboratory staff to: identify any problems in the laboratory, provide insight into the quality of routine laboratory work, promote continuous improvement in performance, inter-laboratory comparability of results and harmonization of methods, and train staff to use EQA information effectively [6,7]. The numerous publications that have emerged on quality management in recent years stress the importance of the EQA schemes and use of the data from EQA surveys for assessing participants’ performance. The design of each scheme differs according to goals chosen, and schemes for the same constituent can, under the management of different organizers, produce varying information on laboratory performance. Such discrepancies are mainly due to the use of different quality specifications. These variations depend, in particular, on difference in methods used for the selection of control materials, statistical procedures (identification of assigned value and outliers), and the assessment of laboratory performance. 1.1. Control materials The EQA scheme organizer is responsible for the appropriate selection of control materials. In compliance with quality standards and scientific recommendations, the materials should simulate fresh human serum as closely as possible in order to avoid interference from the matrix or other components [8 – 14]. The use of human serum poses several difficulties (identification and retrieval of human donors, ethical considerations and consent problems, risk of agents causing infections) and, for practical reasons, commercial control materials (animal sera) are commonly employed. In this sphere, it is essential for the EQA

organizer to be able to demonstrate the quality of materials. In particular, the control materials should be: commutable in order to simulate the measurement process in patient serum as closely as possible (appropriate matrix); sufficiently homogenous (if nonhomogenous material is used, it should be the best available on the market, and the uncertainty of assigned values must be taken into account); sufficiently stable to ensure that they will not undergo any significant change during the entire survey period; and in compliance with all relevant safety standards. 1.2. Statistical procedure An appropriate statistical procedure is a crucial tool for allowing participating laboratories to compare their results with those of other laboratories, obtained on the same sample. Thus, if a patient’s sample is analysed in different laboratories, any variation in results should be such as not to alter the clinical interpretation. Careful considerations should be made concerning the number of results (to allow a statistical significance), identification of method, data classification (all results independently from method, method/ diagnostic system related results), and the procedure used to identify outliers and to estimate the assigned value. The statistical analysis of results must provide reliable estimates of the assigned value (AV) and inter-laboratory variability, data needed for the implementation of proper measures to improve both interlaboratory and inter-method agreement. EQA information can differ depending on the criteria used for data processing, and different techniques can be applied to achieve different objectives and provide different information. The EQA organizer must decide whether to use a parametric or a non-parametric approach, above all analysing the distribution of results and the number of results. The non-parametric method, which is known to be robust, is suggested when the distribution of results is non-gaussian, the number of results is poor, and when a new scheme has been set-up. Therefore, this approach is usually preferred to a parametric evaluation for robustness of median and spread. Usually, in the EQA scheme, the AV is a consensus value, which is the average of results reported by participants in EQA, and is considered an estimation

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

of the ‘‘true value’’, which can be determined only by using definitive methods. When the consensus value is calculated on the basis of all results or is related to each method, it can be influenced by the predominant diagnostic system. In this case, it could be more appropriate to use the consensus value related to each diagnostic system, and the EQA organizers should have to decide when this approach is appropriate (i.e., analyte structure, method principle). However, special attention must be paid to the classification of results in the different groups in order to guarantee homogeneity of data and to facilitate interpretation [8– 14].

89

and inaccuracy provides the TEa to achieve an acceptable result. Numerous quality requirements have been defined over the last 20 years in order to establish analytical quality specifications. In 1999, an International Conference in Stockholm, ‘‘Strategies to Set Global Analytical Quality Specifications in Laboratory Medicine’’, led to a consensus agreement defining a hierarchy of models that should be applied. Where available, and when appropriate for the intended purpose, models higher in the hierarchy must be preferred to those at lower levels [15].

1.3. Assessment of laboratory performance 2. Aim In setting total analytical error (TEa) in EQA schemes, the aim is to improve analytical quality performances within clinical laboratories in order to detect changes in the biological state of patients, a better follow-up of a treatment and an earlier diagnosis. Consequently, acceptability criteria for laboratories’ results in an EQA scheme are of crucial importance: if they are too loose, laboratories with lower performances will not be identified; if they are too stringent, good laboratories will be falsely rejected. The number of ‘‘poor performers’’ in schemes depends on the acceptability criteria used, but it can also be influenced by the statistical procedure used [8 –14]. In choosing acceptability criteria, it must be borne in mind that continuous improvement in laboratory performances is achieved when the total unacceptable performances are limited to a fraction of the overall performances, thereby avoiding to specify limits that are difficult to attain and which discourage laboratories. The laboratory must set the quality specifications including imprecision and inaccuracy goals for each analyte that it uses to evaluate and monitor its work, guarantee its performance and ensure the diagnostic relevance of its results. Each test must satisfy requirements in terms of accuracy and precision and, in every clinical situation, it is necessary to ensure the accuracy required to distinguish between a normal and a pathological result and the precision needed for deciding whether a variation in a result is clinically significant or whether it is due to the analytical variability of the test. The combination of imprecision

In spite of numerous articles in the literature addressing the theoretical and practical bases to design an appropriate scheme, many difficulties are still encountered in routine practice. Each scheme has its own particularities and can emphasize one or more goals. The aim of this paper is to describe the quality specifications used in our EQA schemes in order to design schemes that can assess the reliability of laboratory performance and meet the changing needs and quality recommendations.

3. Material and methods Quality specifications are specific performance standards that should be suggested by scientific societies and expert opinions and be defined on the basis of theoretical and practical considerations [16 –28]. We individuated the quality specifications in the literature and applied them to participating laboratories’ results in EQA schemes managed by the Centre of Biomedical Research (CRB). In particular, we describe the quality specifications applied to: control materials, statistical procedure, and assessment of laboratory performance. 3.1. Control materials Alkaline phosphatase and cholesterol data obtained using commercial control material and human serum controls, in two different EQA surveys (2 –2002 and 1– 2003) are reported.

90

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

3.2. Statistical procedure The non-parametric procedure is used to process EQA results. The assigned value (AV) is the median calculated on method/diagnostic system related results, after the exclusion of values outside the range: median F 3DSrob (outliers), where DSrob is (75jpercentile – 25j percentile)/1.349. The present paper reports the inter-laboratory variability (CV%) data of a diagnostic system obtained for troponin I and the AV data for the CK-MB mass in the 2001 and 2003 cycles concerning the main diagnostic systems. 3.3. Assessment of laboratory performance We identified three different approaches to set analytical goals using the model defined by the International Conference in Stockholm: (a) based on clinical criteria; (b) based on biological variation; (c) resulting from the state-of-the-art [15]. For clinical criteria and biological variation, the goals of imprecision and inaccuracy provided in literature have been used to calculate the TEa on the basis of Fraser’s formula: TEa = 1.65  (0.5CVI ) + 0.25 (CV2I + CVG2)1/2, where CVI is within-subject biological variation and CVG is between-subject biological variation [22,24,26]. For state-of-the-art, on the other hand, the TEa is chosen as the mean of inter-laboratory coefficients of

variation (CVinter%), calculated on homogeneous groups of method/diagnostic systems, after eliminating outliers. We calculated the percentage of acceptable performances (AP) obtained using the TEa set according to the three approaches. The number of acceptable performances is obtained by comparing the bias percentage of each result with TEa. The bias percentage is calculated according to the formula: (EQA result AV)/AV  100. When a bias is lower than TEa the performance is acceptable; when it is higher it is considered unacceptable. The present paper reports data of some analytes with a high clinic-diagnostic value (cholesterol, glucose, and glycated hemoglobin A1c) and analytes characterized by a low biological variation, such as sodium.

4. Results and discussion 4.1. Control materials The results obtained utilizing the 4-nitrophenylphosphatase (4-NPP) method with 2-amino-2-methyl-1-propanol (AMP) buffer or with diethanolamine (DEA) buffer on commercial materials (Fig. 1) and on human serum materials (Fig. 2), are reported. The distribution of results, obtained with the two different buffers, overlaps when commercial materials are

Fig. 1. Results of participating laboratories obtained utilizing commercial control material.

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

91

Fig. 2. Results of participating laboratories obtained utilizing human serum control material.

used and is clearly separated with human serum materials, such as the patients’ samples where the known AMP/DEA ratio is 0.42. In this case, the use of human serum materials is more appropriate because it simulates the behaviour of patient’s serum and it does not present commutability problems. Fig. 3 reports the total cholesterol bias between the median value of a diagnostic system and the median value of cholesterol oxidase/amino-4 antipyrine, peroxidase method (COD-PAP), in two different control

materials (commercial and human serum). Both materials had a negative score, to a lesser extent in the case of human serum materials. When commercial control materials are used in EQA schemes, the diagnostic systems manufacturer involved usually explains this score as a commutability problem. The use of human serum materials confirms that the problem is linked to the lack of standardization and rules out any commutability problem connected with commercial materials.

Fig. 3. Scores between the median of a diagnostic system and the median of the related method, obtained in eight EQA surveys utilizing commercial and human serum control materials.

92

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

The use of commercial control materials is advantageous in view of the large number of analytes and the large range of concentrations they can provide in the same material, good stability during storage and shipping, and lower costs. But a possible conflict of interests with manufacturers’ diagnostic systems and non-commutability attributed to the modification of matrix materials may cause problems. While the use of human serum materials is highly desirable, their use is limited by: high costs, the difficulty in obtaining multiple analytes in large quantities in the same sample, and the difficulty in obtaining the presence of abnormal levels of analytes in the same sample. Therefore, the use of commercial materials in combination with human serum materials appears the best possible solution in EQA schemes. 4.2. Statistical procedure Fig. 4 reports the inter-laboratory variability (CV%) obtained in the most recent EQA cycles for troponin I analysed with a diagnostic system. The increase in CV%, found in the first survey of the 2003 scheme, corresponded to the introduction of a new kit formulation. This kit manufacturer’s, alarmed by this finding, promoted an inquiry in cooperation with the EQA organizers and kit users. Although the inquiry is still in progress, the corrective actions undertaken seem to have already improved the method performance (CV%: from 23.4 to 11.4). The AV of the CK-MB mass obtained on the 2001 and 2003 cycle samples using four different diagnos-

tic systems is reported in Fig. 5. The different diagnostic systems (A –D) give different values and, in some cases, the between-assay differences are about twofold. In the last surveys (2003-4, 2003-5, 2003-6) these differences were slightly reduced. A correct analysis of these data shows that there are standardization problems. The growing number of commercially available test kits lead to values that differ from each other and the incomparability of results obtained with different kits hampers appropriate patient care, if inadequate reference ranges are used. A harmonization of test results must be achieved in different commercially available assays only through cooperation between manufacturers, laboratories and EQA organizers. An adequate statistical procedure allows us to obtain reliable AV and inter-variability coefficients in order to evaluate the variability of results from a method/diagnostic system, and the comparability of results obtained using different methods/diagnostic systems. These data should promote corrective actions designed to achieve agreement between methods and to obtain improvements of each method performance. 4.3. Assessment of laboratory performance The present paper reports, for four analytes, the percentage of acceptable performances obtained in the EQA results from the 2003 cycle by means of the application of goals based on: clinical criteria, biolog-

Fig. 4. Inter-laboratory variability (CV%) obtained in the most recent cycles for troponin I analysed with a diagnostic system.

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

93

Fig. 5. Assigned values of CK-MB mass obtained in the 2001and 2003 cycles samples involving four different diagnostic systems (A, B, C, D).

ical variation, and state-of-the-art. The TEa used in our EQA schemes, are also reported. We chose the specifications based on biological variation and compared them with the state-of-the-art, which seemed to be the most practical approach because clinical data are not always available. If the limits based on biological variation were too narrow (a high percent-

age of laboratories having poor performances) or too wide (a high percentage of laboratories having good performances) we replaced them with other limits that are multiples or under-multiples of those derived from Fraser’s formula, so that no more than 25% of the results of laboratories considered had an unacceptable performance [27].

94

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

Table 1 Total percentage of acceptable performances obtained in the 2003 cycle Total performances

Cholesterol Glucose Sodium Glycated hemoglobin

Acceptable performances Clinical goal

Biological goal

State-of-the-art goal

EQA goal

N

N

(%)

N

(%)

N

(%)

N

(%)

4130 4190 4031 1710

3737 3272

90.5 78.1

1338

78.2

3820 3771 1814 664

92.5 90.0 45.0 38.8

2522 2472 2745 1030

61.1 59.0 68.1 60.2

3453 3313 3048 1305

83.6 79.1 75.6 76.3

4.4. Cholesterol and glucose For cholesterol and glucose, most of the results attained the analytical goals based on clinical criteria (cholesterol: 90.5%; glucose: 78.1%); a higher percentage of results the biological goals (cholesterol: 92.5%; glucose: 90.0%). Otherwise, about 60% (cholesterol: 61.1%; glucose: 59%) of results were acceptable when goals based on state-of-the-art were applied. Our EQA goals, which are narrower goals in order to encourage laboratories to achieve quality improvement, are attained by 83.6% (cholesterol) and 79.1% (glucose) of results. Although current technology should allow the use of more restrictive goals (mean CV% = 3.3% and 3.33%, for cholesterol and glucose, respectively), their advantages do not justify the efforts to maintain this quality level. Moreover, when TEa = 3.3 and

TEa = 3.33 are used, the percentage of unacceptable results is too high (about 40%) and could discourage laboratories from attempting to improve upon quality (Table 1 and Figs. 6, 7). 4.5. Sodium The clinical goal for sodium is not available and only 45% of results achieve the biological goal because the biological variation of sodium is very low. The percentage of acceptable results is 68.1% when the goal is based on state-of-the-art. The goal used in our EQA scheme seems to be the more appropriate one because it takes into consideration current methodological possibilities and only a low percentage of laboratories have unacceptable performances. The manufacturers of diagnostic systems should improve upon the precision and accuracy of

Fig. 6. Percentage of acceptable performances obtained in eight different control materials on applying the different goals. The data are shown with the sequential distribution of the 2003 cycle and the median values obtained with cholesterol oxidase, amino-4 antipyrine, peroxidase method (CHOD, PAP) ranging from 3.94 mmol/L to 5.00 mmol/L.

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

95

Fig. 7. Percentage of acceptable performances obtained in eight different control materials on applying the different goals. The data are shown with the sequential distribution in 2003 cycle and the median values obtained with glucose oxidase, peroxidase method (GOD, POD) ranging from 4.60 mmol/L to 6.38 mmol/L.

their methods to enable laboratories to achieve the biological goal (Fig. 8). 4.6. Glycated hemoglobin In the case of glycated hemoglobin, the AV is determined using the DCCT method, which provides a more reliable evaluation of performances when the consensus value is used. The biological goal is attained by a low percentage of results (38.8%) [26]. The percentage of laboratories that attained the other analytical goals ranged from

60.2% to 78.2% for all goals applied. The use of the value obtained with the DCCT method (as AV), chosen by a group of experts according to scientific recommendation, aims to stimulate the improvement of standardization between diagnostic systems and the agreement between laboratories results (Fig. 9). The application of EQA goals, chosen using the above described criteria, was found to be effective. In fact, the laboratories’ performances significantly improved over time. For example, in the cardiac markers scheme, the number of unacceptable performances, calculated on results obtained in the same laboratories,

Fig. 8. Percentage of acceptable performances obtained in eight different control materials on applying the different goals. The data are shown with the sequential distribution in 2003 cycle and the median values obtained with indirect potentiometric method (ISE) ranging from 141.0 mmol/L to 145.0 mmol/L.

96

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

Fig. 9. Percentage of acceptable performances obtained in eight different control materials on applying the different goals. The data are shown with the sequential distribution in 2003 cycle and the median values obtained with DCCT method ranging from 5.1% to 10.2%.

decreased (from 11.6% to 5.3%) and the number of optimum performances increased (from about 49% to about 60%) for troponin I. Similarly, unacceptable performances decreased for myoglobin (from 19.5% to 11.6%), and for the CK-MB mass (from 13.2% to 3.9%).

organizers and manufacturers of diagnostic systems is crucial to establishing the most adequate possible quality specifications, which comply with criteria for efficiency and effectiveness.

References 5. Conclusion The increased diagnostic value of the test results and their recognized impact on clinical outcome stress the importance of the laboratory test. Therefore, appropriate quality specifications should be established for each test in order to guarantee high-quality analytical performances. EQA is an important component in quality management, providing comparisons with other laboratories and with established quality specifications. EQA schemes should be integrated in the laboratory process, as this will guarantee that the following are achieved: problems are focused on, the quality control system is constantly reviewed and updated, and efforts to bring about improvement are supported. To design reliable EQA schemes, it is very important to define the quality specifications required to promote quality improvement in laboratories. These must comply with quality requirements and scientific recommendations and their appropriateness enhances the value of laboratory test results and clinical practice, and can be effective in evaluating the patient’s outcome. Cooperation between laboratories, EQA

[1] Wong ET. Improving laboratory testing: can we get physicians to focus on outcome? Clin Chem 1995;41(8(B)):1241 – 7. [2] Plebani M. The clinical importance of laboratory reasoning. Clin Chim Acta 1999;280:35 – 45. [3] Plebani M. Appropriateness in programs for continuous quality improvement in clinical laboratories. Clin Chim Acta 2003;333:131 – 9. [4] Waise A. Clinical audit and the contribution of the laboratory to clinical outcome. Clin Chim Acta 1999;280:47 – 57. [5] Lewis LS. Quality control: good laboratory practice—an overview of quality assurance. www.who.int. [6] Libeer JC. External quality assurance programmes in medical laboratories. Accredit Qual Assur 2001;6:151 – 3. [7] Libeer JC. Role of external quality assurance schemes in assessing and improving quality in medical laboratories. Clin Chim Acta 2001;309(2):173 – 7. [8] Libeer JC. External quality assessment in clinical laboratories. European perspectives: today and tomorrow. Antwerpen: Universitaire Instelling Antwerpen; 1993. [9] International Federation of Clinical Chemistry (IFCC). Fundamental for External Quality Assessment (EQA) 1997. [10] ISO Guide 43. Proficiency testing by inter-laboratory comparison. Second ed. International Organization for Standardization (ISO); 1997. [11] NCCLS. Using proficiency testing (PT) to improve the clinical laboratory. Approved guideline GP27-A. 1999;19(15). [12] ILAC-G13. Guidelines for the requirements for the competence of providers of proficiency testing schemes. 2000. [13] Clinical Pathology Accreditation (UK). Standards for EQA

L. Sciacovelli et al. / Clinica Chimica Acta 346 (2004) 87–97

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

Schemes in Laboratory Medicine. Version 4.01 (September 2003). NCCLS Using Proficiency Testing (PT) to improve the Clinical Laboratory. Approved Guideline, GP27-A. (August 1999);19(15). Petersen HP, Fraser CG, Kallener A, Kenny D. Strategies to set global analytical quality specifications in laboratory medicine. Scand J Clin Lab Invest 1999;59(7):475 – 585. Petersen HP, Groth T, de Verdier CH. Principles for assessing analytical quality specifications and their use in design of control systems. Uppsala J Med Sci 1993;98:195 – 214. Fraser CG, Petersen PH. Desirable standards for laboratory tests if they are to fulfill medical needs. Clin Chem 1993; 39/7:1447 – 55. Fraser CG, Petersen HP. Quality goals in external quality assessment are best based on biology. Scand J Clin Lab Invest 1993;53(Suppl. 212):8 – 9. Petersen HP, Blaabjerg O, Irjala K, Ice`n A, Bjotro K. Elements of analytical quality. Helsinki, Finland: Nordkem Clinical Chemistry Project, 1994. p. 187 – 94. Stockl D, Baadenhuisjen H, Fraser CG, Libeer J-C, Petersen HP, Ricos C. Desirable routine analytical goals for quantities assayed in serum. Eur J Clin Chem Clin Biochem 1995;33: 157 – 69. Fraser CG, Hyltoft Petersen P, Libeer J-C, Ricos C. Proposal for setting generally applicable quality goals solely based on biology. Ann Clin Biochem 1997;34:1 – 8.

97

[22] Sebastia`n-Ga`mbaro MA, Liron-hernandez FJ, Fuerte-Ardeiu X. Intra- and inter-individual biological variability data bank. Eur J Clin Chem Clin Biochem 1997;35(11):845 – 52. [23] Fraser CG. General strategies to set quality specifications for reliability performance characteristics. Scand J Clin Lab Invest 1999;59:487 – 90. [24] Rico`s C, Alvarez V, Cava F, Garcia-Lario JV, Hernandez A, Jimenez CV, Minchinela J, Perich C, Simon M. Current databases on biological variation: pros, cons and progress. Scand J Clin Lab Invest 1999;59:491 – 500. [25] Westgard JO. The need for a system of quality standards for modern quality management. Scand J Clin Lab Invest 1999; 59:483 – 6. [26] Garde AE, Hansen AM, Skovgaard LT, Christensen JM. Seasonal and biological variation of blood concentrations of total cholesterol, dehydroepiandrosterone sulfate, hemoglobin A1C, IgA, prolactin, and free testosterone in healthy women. Clin Chem 2000;46(4):551 – 9. [27] Sciacovelli L, Secchiero S, Zardo L, Plebani M. External quality assessment schemes: need for recognised requirements. Clin Chim Acta 2001;309:183 – 99. [28] Secchiero S, Sciacovelli L, Zardo L, Plebani M. Appropriateness of cholesterol and triglycerides reporting checked by external quality assessment programs. Clin Chim Acta 2003; 333(2):221 – 30.