Method Selection and Evaluation D Kefaya EL Sayed

Method Selection and Evaluation D. Kefaya EL- Sayed Mohamed Prof. Of Clinical Pathology (Clinical Chemistry), Mansoura University

Before a new test or methodogy is introduced into the laboratory. both managerial and techinal information must be compiled and carefully considered. The information should be collected from many different sources, including manufacturer sales representatives, colleagues, scientific presentations, and the scientific literature.

Linear range:

O. D 1. 1 1. 0 0. 9 0. 8 0. 7 0. 6 0. 5 0. 4 0. 3 0. 2 0. 1 0. 0 200 400 600 800 Glucose concentration (mg/dl)

Observed value True value RE X µ SE TE

Inaccuracy: Ø The difference between a measured value and its true value. Ø Is due to the presence of systematic analytical error (S. E. ) Ø S. E. can be constant or proportional.

Test Values 30 Constant E Ideal Proprtional E 0 30 Reference Values n

Inaccuracy Ø Can be estimated from three studies: n recovery. n interference. n a comparison-of-methods study

Example of a recovery study Sample preparation Sample 1: 2. 0 ml serum + 0. 1 ml H 2 O Sample 2: 2. 0 ml serum +0. 1 ml 20 mg /dl calcium standard Sample 3: 2. 0 ml serum +0. 1 ml 50 mg /dl calcium standard Concentration Calcium measured Sample 1 7. 50 mg/dl Sample 2 8. 35 mg/dl Sample 3 9. 79 mg/dl added 0. 95 mg/dl 2. 38 mg/dl Recovered Recovery 0. 85 mg/dl 2. 29 mg/dl 89% 96%

Calculation of recovery Concentration added = standard concentration x Ml standard + ml serum = 20 x 0. 1 = 20 x 1 = 20 = 0. 95 (Sample 2) 2+0. 1 2. 1 x 10 21 conc. recovered = conc. ( diluted test) – conc. ( baseline) = 8. 35 -7. 5=0. 85 recovery = Conc. Recovered = x 100% Conc. added = 0. 85 x 100 = 89 % ( sample 2) 0. 95

( b )Interference: Ø The interference experiment is used to measure systematic errors caused by substances other than the analyte

Interference: Ø n n An interfering material can cause systematic errors in one of two ways : The material itself may react with the analytical reagents. Or it may alter the reaction between the analyte and the analytical reagents

Ø Ø Ø Interference: The interference likes recovery except that the substance suspected of interference is added to the patient sample. The concentration of the potentially interfering material should be in the maximally elevated range. If an effect is observed its concentration should be lowered to discover the concentration at which test resultes are frist invalidated

Example of an Interference study Sample preparation Sample 1 : 1. 0 ml serum + 0. 1 ml H 2 O (base line) Sample 2 : 1. 0 ml serum + 0. 1 ml of 10 mg/dl magnesium standard Sample 3 : 1. 0 ml serum + 0. 1 ml of 20 mg/dl magnesium standard Sample 1 Sample 2 Calcium Measured 9. 80 mg/dl 10. 53 mg/dl 11. 48 mg/dl Magnesium added 0. 91 mg/dl 1. 81 mg/dl Interference 0. 73 mg/dl 1. 68 mg/dl

Calculation of interference Concentration added = standard concentration x ml standard + ml serum = 10 x 0. 1 = 10 = 0. . 91 mg/dl (sample 2) 1. 1 11 Interference = conc. ( diluted ) – conc. ( baseline) = 10. 53 – 9. 8 =0. 73 (sample 2)

Ø Ø Common interferences (eg. , hemoglobin, lipids, bilirubin, anticoagulants preservatives, and so on)also should be tested. Glick and Ryder have presented " interferographs " for various chemistry instruments – these are graphs relating analyte concentration measured versus interferent concentration

( C ) comparison – of methouds Experiment: Ø The best comparative method that can be used is the reference method , which is a method with negligible inaccuracy in comparison with its imprecision.

Ø Reference methods may be laborious and timeconsuming Because most laboratories are not staffed and equipped to perform reference methods, the results of the test method are usually compared with those of the method routinely in use.

Ø n n n Westgard et al 27 and the NCCLS 35 recommend that at least 40 sample and preferably 100 samples, be run by botli methods. span the clinical range represent many different pathologic conditions. Duplicate analyses of each sample by each method.

Ø Ø If 40 specimens are compared , two to five patient specimens should be analyzed daily for a minimum of 8 days. If 100 specimens are compared the comparison study should be carried out during the 20 -day replication study.

Test method Regression line Reference n

Example of Method selection and evaluation (Glucose in Serum) Analytical Needs : - Ø Rapid procedure is needed at time out side the regular working hours. n Sample Volume of 0. 2 ml or less n Analytical range of 0. 0 to 500. 0 mg/dl. n Turn arround time 30 min. or less is needed n

Ø Quality goals: - The medical decision levels of interest are: n 50 (hypoglycemia) n 200 mg/dl (hyperglycemia) n The decision level for screening is 140 mg/dl is not needed. Precision goals are 1. 5 mg/dl at 50 mg/dl 5. 0 mg/dl at 200 mg/dl Total error goals (TEA) are 6. 0 mg/dl at 50. 0 mg/dl 20. 0 mg/dl at 200. 0 mg/dl

Ø Method selection : - § In kit form Can be set up on an existing laboratory instrument Ø Requirements: - § § Primary reference solutions are prepared for calibration Control materials Randomly selected clinical specimen.

§ § Within – Run Precision : Analyze 20 aliquots of low abnormal control mean ±SD 56. 5 mg/dl ± 0. 7 Analyze 20 aliquots of moderatly high control mean ± SD 182. 6 ± 2. 1 mg/dl SDs are less than the allowable error in non automated procedure.

Ø Day to Day precision: § Two Q. C pools are analyzed for 20 days mean ± SD § SD : 1. 5 mg /50 mg glucose 5. 0 mg/200 mg glucose

Ø § § Analytical Range : Glucose solutions prepared from a stock glucose reference solution of 1000 mg/dl Duplicate analyses The plot shows excellent linearity to 600 mg/dl. which meets the specification for linearity to 500 mg/dl. The absorbance at zero glucose concentration is the reagent blank value

O. D 1. 1 1. 0 0. 9 0. 8 0. 7 0. 6 0. 5 0. 4 0. 3 0. 2 0. 1 0. 0 200 400 600 800 Glucose concentration (mg/dl)

Ø § § § Recovery: Two pooled sera are perpared as baseline specimens 61 mg/dl , 171 mg/dl concentrated glucose solution is prepared (10, 000 mg/dl) two different amounts of glucose are added to 9. 6 ml of pooled sera

Recovery: - Glucese solution 10, 000 mg/dl 0. 0 ul 100. 0 ul 400. 0 ul Nacl 0. 15 mol/l 400. 0 ul 300. 0 ul Total 61 mg/dl 171 mg/dl 10. 0 m. L a a a b b b

§ § Quadruplicate assays of A and B are performed and take the mean Subtract the original (basal) glucose levels (61, 171) from the mean of each quadruplicate Recovered glucose § Recovery % = § Recovered glucose Glucose added X 100 Proportional error = 100 - recovery%

Recovery: - Glucose Gluco Pool Gluc added se e o Re d se cov s fo mg/ ere ml/10 e u d d ml r n l mg a d /dl A A A B B B 0. 0 100 400 61 159 457 171 268 562 --15961 =9 8 369 --97

Interference: § The potential interference of visible abnormal specimens is tested by : A analyzing a series of icteric , turbid, and hemolyzed sera using the new method an established method know to be free of such interferenc Ø

§ The difference ( ) the two methods are 1 – 4 mg which is less than the allowable TEA. B add a small amount of highly concentrated solutions to pooled sera e. g. Ascorbic acid , sod. Salicylate, Trisodium citrate, Heparin, disod. EDTA. § The difference must be < TEA.

Interference: INTERFERENCE Type Concentr ( Ser ation um) METHOD FOR GLUCOSE Compar ative (mg/d l) Norma l Bil. (Ict eric ) Hb. (H em o. ) Test 113± 5. 6 --- 112± 4. 4 6. 5 mg/dl 98± 5. 3 96 mg/dl 122± 3. 2 97± 6. 4 120± 7. 6