Internal Quality Control QC for Medical Laboratories An

Internal Quality Control (QC) for Medical Laboratories: An introduction Dr. Otto Panagiotakis and Dr. Alexander Haliassos ESEAP – Greek Proficiency Testing Scheme for Clinical Laboratories http: //www. eseap. gr info@eseap. gr

• Analytical Quality Control • The most important tool for ensuring the quality of laboratory results through the identification and reduction of errors It includes two parallel mechanisms: • Internal (intra-laboratory) quality control • External quality control, or External quality assessment (EQA) or Proficiency Testing (PT)

What means a value of total cholesterol 245 mg/d. L reported for the analysis of an QC control? Correct result: There is not a value, but a range from repeated measurements We need a tool to compare the reported with the expected 235 245 255 265 240 250 260 result. It is the control chart Levey-Jennings

Control charts Levey-Jennings: +3 s +2 s +1 s mean -1 s -2 s -3 s Central horizontal line: expected mean value Dotted horizontal lines: control limits (mean ± n. SD)

Control charts Levey-Jennings: Their design is based on the assumption that: • The values resulting from previous measurements are subject to random variation • This variation follows a uniform (normal) distribution

How we draw the control charts? • We select a parameter (p. ex. Cholesterol) • we measure this parameter in a control material for 20 days using the method (assay) and the instrument (analyzer) that we evaluate • from those values we calculate the mean and the SD p. ex. mean = 200 mg/d. L and SD = 4 mg/d. L • subsequently, the control limits at the level of 2 s, 3 s mean ± 2 s: 200 ± 2(4)=200 ± 8 � from 192 to 208 mean ± 3 s: 200 ± 3(4)=200 ± 12 � from 188 to 212

Cholesterol (mg/d. L), Lot No: xxx, January 2015 216 x+3 s 212 x+2 s 208 mean 200 192 x-2 s 188 184 x-3 s 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

Interpretation of control charts (1) • The control charts reflect the distribution resulting from the initial measurements of the control materials • In the charts we draw the values of each measurement • Evaluation: depending on their position in the chart 1) If the analytical procedure is correct, the new measurements will have the same distribution with the originals: • Extremely rare (0, 3%) a value> mean ± 3 s • Unlikely (5%) a value> mean ± 2 s • Very likely (32%) a value> mean ± 1 s (limits without a value)

Control Rules to decide whether a series of measurments is under control or out of control Control rules control limits 12 s mean ± 2 s 13 s mean ± 3 s Single rule methods Multiple rules methods

Interpretation of control charts (2) 2) If the analytical procedure has a problem, it increases the probability of a value exceeding the control limits This can happen : • Either with the appearance of a constant error (bias) (shifting of the mean of the distribution to or values) • or by increasing the random error (enlargement of the distribution) Situation out of control / Unacceptable results

Situation under control (normal) • Systematic error (bias) Distribution enlargement

The problem of erroneous rejects The QC methods are detection systems The detection systems have some characteristics: The frequency of true warnings, true alarms The frequency of erroneous warnings, false alarms In the QC methods they are called respectively : • Error detection probability (Ρ) • Probability of false rejection (Ρο) The ideal on a single rule QC method would be: Ρ=1, 0 (100%) and Ρο=0 (0%) However, for each control rule: Ρ < 1 and Ρο > 0 A realistic goal is : Ρ=0, 90 and Ρο=0, 05

Rule 12 s: high Ρ and high Ρο For Ν=1 Ρο=5% For Ν=2 Ρο=9% For Ν=3 Ρο=14% • The single rule QC method using the 12 s rule should only be used with Ν=1. • If Ν=2, almost a false rejection in 10 • This is not a problem of the analytical procedure. • This is an intrinsic problem of the QC method and related to the selected control threshold.

Rule 13 s: low Ρο and low Ρ resulting in non detected large errors As the control limits are extended erroneous rejections (Po) decrease, but also P decreases If a rule with high P is selected will have also high Ρο Single rule QC methods have serious drawbacks Need to find other QC methods

Multiple rules methods They do not use a single control rule but a combination of rules (at least 2) Advantage: Low Po and simultaneously high P The most well known Westgard method: • 6 control rules • 2 control sera (N=2) resulting to • 2 Levey-Jennings control charts L-J, one for each serum (control material) • Control limits at three levels(± 1 s, ± 2 s, ± 3 s)

Cholesterol (mg/d. L), Lot No: xxx, January 2015 216 x+3 s 212 x+2 s 208 x+1 s 204 200 196 x-1 s 192 x-2 s 188 184 x-3 s 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

The 6 control rules according to Westgard 12 S 13 S 22 S R 4 S 41 S 10 x

Rule 12 S One value (measurement) > 2 s limit +3 s +2 s +1 s mean - 1 s - 2 s - 3 s 1 2 3 4 5 6 7 8 9 10 Not a rejection but warning for a potential problem Further control is required based on the other criteria

Rule 13 S One value (measurement) > 3 s limit +3 s +2 s +1 s mean - 1 s - 2 s - 3 s 1 2 3 4 5 6 7 8 9 10 Applied in a series for each of the two sera Sensitive to random errors

Rule 22 S 2 consequent values > the same limit of 2 s +3 s +2 s +1 s mean - 1 s - 2 s - 3 s 1 2 3 4 5 6 7 8 9 10 In the last 2 series for each serum In the same series for both

+3 s Rule 22 S +2 s 2 consequent values > the same limit of 2 s +1 s mean +3 s - 1 s +2 s - 2 s +1 s - 3 s mean 1 - 1 s - 2 s +3 s - 3 s +2 s 1 2 3 4 5 6 7 8 9 10 +1 s mean In the last 2 series for each serum- 1 s - 2 s In the same series for both - 3 s Sensitive to systematic errors 1 2 3 4 5 6 7 8 9 10

Rule R 4 S The difference in value of the two sera> 4 s +3 s +2 s +1 s mean - 1 s - 2 s - 3 s 1 2 3 4 5 6 7 8 9 10 In the last 2 series for each serum In the same series for both

+3 s Rule R 4 S +2 s The difference in value of the two sera> 4 s +1 s mean +3 s - 1 s +2 s - 2 s +1 s - 3 s mean 1 - 1 s - 2 s +3 s - 3 s +2 s 1 2 3 4 5 6 7 8 9 10 +1 s mean In the last 2 series for each serum- 1 s - 2 s In the same series for both - 3 s Sensitive to systematic error 1 2 3 4 5 6 7 8 9 10

Rule 41 S 4 consecutive values > the same limit of 1 s +3 s +2 s +1 s mean - 1 s - 2 s - 3 s 1 2 3 4 5 6 7 8 9 10 In the last 4 series for each serum In the last 2 series for both

+3 s Rule 41 S +2 s 4 consecutive values > the same limit of 1 s +1 s mean +3 s - 1 s +2 s - 2 s +1 s - 3 s mean 1 - 1 s - 2 s +3 s - 3 s +2 s 1 2 3 4 5 6 7 8 9 10 +1 s mean In the last 4 series for each serum- 1 s - 2 s In the last 2 series for both - 3 s Sensitive to systematic errors 1 2 3 4 5 6 7 8 9 10

Rule 10Χ 10 consecutive values at the same side of the mean +3 s +2 s +1 s mean - 1 s - 2 s - 3 s 1 2 3 4 5 6 7 8 9 10 In the last 10 series for each serum In the last 5 series for both

Rule 10Χ 10 consecutive values at the same side of the mean +3 s +2 s +1 s mean - 1 s - 2 s - 3 s 1 +3 s +2 s 1 2 3 4 5 6 7 8 9 10 +1 s mean In the last 10 series for each serum - 1 s - 2 s In the last 5 series for both - 3 s Sensitive to systematic errors 1 2 3 4 5 6 7 8 9 10

Flowchart according to Westgard values control 12 s no Situation under control - Series Accepted no yes 13 s yes no 22 s yes no R 4 s yes no 41 s no yes Situation out of control - Series Rejected 10 x yes

Differences of Internal and External QC Internal QC • performed daily in the laboratory • uses samples (control materials) of known concentration • it is always required External QC • performed periodically (weekly, monthly …) • uses samples (control materials) of unknown concentration • useful in conjunction with the internal QC

External QC does not replace the Internal QC • the inter-laboratory comparisons are infrequent • the results are reported deferred (not in real time) and therefore it is not possible the immediate intervention with corrective measures • even if the performance is satisfactory, it assures the proper functioning of the laboratory only on the day of the inspection (participation) These programs do not lead to quality improvement of the laboratory if it is not performed daily the internal QC.