Proficiency Testing
CLIA 88 requires that laboratories successfully participate in an approved proficiency testing program. Proficiency testing is the external component of quality control since it involves peer review. It primarily measures accuracy because each laboratory’s results are compared to the interlaboratory mean, which is assumed to be the true value. Each individual laboratory’s accuracy is evaluated by how close its results are to the mean.Proficiency test results are reported as standard deviation indexes (SDIs). SDI is calculated by the following formula:
SDI = (your result - interlaboratory mean)/ interlaboratory SD.
This index represents the number of standard deviations each result is from the mean.For example, if your laboratory gets a result of 40 mg/dL, the interlaboratory mean is 36 mg/dL and the interlaboratory standard deviation is 2 mg/dL, then the SDI is (40 - 36)/2 = 2. If the SDI cutoff is +/-2 SDI, then your laboratory has borderline acceptability. If the interlaboratory standard deviation is 1.5 mg/dL, instead of 2 mg/dL, then the same laboratory would obtain an SDI of (40 - 36)/1.5 = 2.67. If the interlaboratory standard deviation is 3.0 mg/dL, the laboratory’s SDI would be (40 - 36)/3 = 1.33. Thus, the SDI method of evaluation is most dependent on the relative performance of peer laboratories. If proficiency testing data have a Gaussian distribution, then 68% of results should fall within 1 SDI, 95% within 2 SDI, and 99.7% within 3 SDI. A laboratory should strive to have its results fall within the middle of its peer group.
The interlaboratory mean can be calculated from the laboratory’s peer group or the all method mean. If a laboratory is using the manufacturer’s calibrators and operating the instrument as directed by that company, it should compare itself against its peer group. However, to be valid peer groups should contain at least 20 participants. If peer groups of sufficient size do not exist, laboratory results should be compared against the all method mean.
Quantitative analytes are graded on the basis of HCFA limits, some of which are quite broad. Pathologists and laboratory supervisors should look at the values in the following tables, divide each HCFA performance limit by three, and compare the results to their routine day to day precision (CVs). If all CVs are less than one third of the HCFA limits there is essentially zero risk of failing a proficiency test event. If CVs are higher than one third of the HCFA limits, random error is very likely to cause a proficiency test result to fall outside the proficiency test limit. For those tests where internal CVs exceed one third of the HCFA limits, troubleshooting should be undertaken to reduce imprecision. Appendix B contains a more complete list, including medical decision levels.
HCFA Proficiency Test Limits for Acceptable Performance
Hematology Analyte |
Target Value +/- |
Hemoglobin |
7% |
Hematocrit |
6% |
RBC |
6% |
White blood cell count |
15% |
Platelet count |
25% |
Prothrombin time |
15% |
APTT |
15% |
Fibrinogen |
20% |
Toxicology |
Target Value + or - |
Alcohol, blood |
25% |
Blood Lead |
10% or 4 ug/dL |
Carbamazepine |
25% |
Digoxin |
20% or 0.2 ng/mL |
Ethosuximide |
20% |
Gentamicin |
25% |
Lithium |
0.3 mmol/L or 20% |
Phenobarbital |
20% |
Phenytoin |
25% |
Primidone |
25% |
Procainamide |
25% |
Quinidine |
25% |
Theophylline |
25% |
Valproic acid |
25% |
Routine Chemistry |
Target Value +/- |
Albumin |
10% |
ALP |
30% |
ALT |
20% |
Amylase |
30% |
AST |
20% |
Bilirubin, total |
0.4 mg/dL or 20% |
Calcium, total |
1.0 mg/dL |
Chloride |
5% |
Cholesterol, HDL |
30% |
Cholesterol, total |
10% |
CK |
30% |
CK isoenzymes |
3 SD or MB elevated (+ or -) |
Creatinine |
0.3 mg/dL or 15% |
Glucose |
6 mg/dL or 10% |
Iron |
20% |
LD |
20% |
LD isoenzymes |
30% or LD1/LD2 |
Magnesium |
25% |
pCO2 |
5 mm Hg or 8% |
pH |
0.04 units |
pO2 |
3 SD |
Potassium |
0.5 mmol/L |
Sodium |
4 mmol/L |
Total Protein |
10% |
Triglycerides |
25% |
Urea nitrogen |
2 mg/dL or 9% |
Uric acid |
17% |
Endocrinology |
Target Value + or - |
Cortisol |
25% |
Free Thyroxine |
3 SD |
hCG |
3 SD |
T3 Uptake |
3 SD |
Triiodothyronine |
3 SD |
TSH |
3 SD |
Thyroxine |
20% or 1 ug/dL |
Proficiency test results must be reviewed by the laboratory director and section supervisors.They should be examined for unacceptable results, unusual trends, sudden shifts, or other problems. The best way to detect problems is to examine SDI results. The following rules are useful in evaluating SDI results:
- Do 2 or more of the 5 SDI results exceed +/- 1? If not, a significant error is unlikely and evaluation is complete. If so, further questions need to be asked.
- Is the average of the 5 SDI results more than +/- 1.5? If so, significant bias (systematic error) is present and calibration data should be reviewed to determine if a shift has occurred. Bias can usually be eliminated by recalibration.
- Is one result more than +/- 3 SDI ? If so, there is a high probability of random error.
- Does the range of SDI between the largest and smallest PT result exceed 4 SDI? If so, random error is a possibility and the procedure should be evaluated for potential sources of imprecision.
Comparison of SDI values during successive surveys is also useful. If a laboratory is consistently below the mean, the SDI will be a series of negative numbers. If the laboratory is consistently above the mean, the SDI will be a series of positive values.
If an analyte falls outside of the acceptable range it will be flagged. All proficiency test errors must be thoroughly investigated and documented. Investigation should include the following steps.
- Review your copy of the original PT answer sheet and look for transcription errors, transposition of answers, coding errors, miscalculations, or grading mistakes.
- Review all test result printouts - was the result accepted in a nonlinear range?
- Review quality control results from the same day - was the PT result accepted when QC was out of control? Were there any QC shifts or trends?
- Retest the original PT sample if possible. Iif repeat testing produces an acceptable result, then the original result was affected by analytical error. If the repeat result is still unacceptable, attention should be focused on specimen identification and labeling errors. Specimen preparation should also be reviewed for reconstitution or dilution error, pipetting problems, and aliquot evaporation due to delay between reconstitution and analysis. Survey material should be examined for instability, contamination, chemical interference, an incompatible matrix effect, or shipping delays.
- Review previous survey to determine if this analyte was within acceptable range.