Page 22 - qp_17

APPENDIX 4

LIMIT OF DETECTION & QUANTIFICATION STUDY

Limit of detection is applicable to tests that are normally present in low concentrations in human
body. Small changes in the concentration may indicate the presence of disease.

For the purpose of this study, the instrument should produce the absolute value.

A) Verification of limit of blank (LoB) provided by the manufacturer
1. Run at least 20 replicates of blank material (as recommended by manufacturer), record
the data.
2. Observe the data, if at least 85% of 20 replicates (at least 17 replicates) show value less
than LoB provided by the manufacturer, the limit of blank provided by the manufacturer
can be used, if not, go to the next step.
3. Establish 95% confidence limit (one side confidence) by calculating mean blank +
1.65SD of blank.
4. Proceed to verification of limit of detection.

B) Verification of claimed limit of detection (LoD)
1. Use patient’s samples with concentration in the lower range (eg suppressed TSH in
patients with thyrotoxicosis). Use of multiple samples is preferred than pooled sample.
2. Run a minimum of 20 replicates over several days.
3. Record the data as in table below. Arrange the data from low value to high value in
order.

Table 2 : Data records for LoD
Ranks Blank values Sample values
1
2
3
…
20

4. Calculate the 95th percentile of the blank value using the following formula:
LoB result at position= [n (p/100) + 0.5] = (n x 0.95) + 0.5, for 95% Confidence Interval
where n = number of run. Take or calculate LoB value at position calculated above.

5. Check value of sample value in the table above, if more than 85% of the value exceed
theblank value at 95th percentile then LoD claimed by the manufacturer is acceptable. If
not,establish laboratory LoD using the formula: LoD = LoB + 1.645 (SD low
concentration sample)

C) Verification of limit of quantification (LoQ)

Limit of quantitation is the lowest actual amount of an analyte that can be reliably detected
(LoD) and that can be quantitatively determined with stated acceptable precision and trueness,
that is bias plus 2 times method SD satisfy the goal of Total Error. (Tea = bias + 2SD). LOQ =
lowest actual amount of an analyte that can give total error within the specification.
LOQ is unlikely to be claimed by the manufacturer due to issue in calculation bias. Therefore
this guideline will only focus on functional sensitivity that is based on CV criteria.