Standard Deviation Index Calculator (SDI for proficiency testing & peer comparison)

Understanding the standard deviation index calculator — concept and principle

This calculator computes the standard deviation index (SDI) for laboratory proficiency testing or inter-laboratory quality control: how far your result is from the peer-group mean, measured in group standard deviations. For several samples it gives each SDI, the average, the largest, the range and checks against common interpretation rules. Use it when reviewing PT or external quality assessment reports.

Four samples sit within ±1 SD of the group; one at +1.5 is worth checking.

How it is calculated

SDI = (your result − group mean) ÷ group SD

For a set of samples it also reports:

Average SDI and largest |SDI|.
How many samples are beyond ±1.0.
Range of SDIs (highest − lowest).

Interpretation guidance applied to the size of the SDI: up to 1.0 acceptable; 1.0–1.9 investigate for bias; 2.0 or more among the poorest performers. For a multi-sample survey: two or more SDIs beyond 1.0 warrant investigation; an average beyond 1.5 suggests systematic bias; a single SDI beyond 3.0 suggests random error; a range over 4 SDI suggests random error.

The principle

An SDI is a z-score against your peer group. Dividing by the group SD puts all analytes and concentration levels on one scale, so a difference of 5 units on a test with a wide spread can be compared with a difference of 0.5 on a test with a narrow spread. A consistent sign across samples, or an average far from 0, points to a systematic bias such as calibration. Large single deviations or a wide spread between samples point to random error such as sample handling or transcription mistakes.

Worked example

Default samples (result, mean, SD): (102, 100, 4), (97, 100, 4), (108, 103, 5), (95.5, 98, 3), (124, 115, 6).

SDIs: 2 ÷ 4 = 0.50; −3 ÷ 4 = −0.75; 5 ÷ 5 = 1.00; −2.5 ÷ 3 ≈ −0.83; 9 ÷ 6 = 1.50.
Average SDI = (0.50 − 0.75 + 1.00 − 0.83 + 1.50) ÷ 5 ≈ 0.28.
Largest |SDI| = 1.5: investigate for bias (1.0–2.0).
One sample beyond ±1.0; range = 1.50 − (−0.83) ≈ 2.33.
None of the multi-sample rules is triggered.

How to read the result

Where does your result fall?

An average near 0 means no overall bias. Look at single samples too: here one result sits 1.5 SD above the group, which is worth a look even though the overall picture is within guidelines. Track SDIs over several surveys; a trend in one direction is more informative than one value.

Common mistakes

Using your own laboratory's SD instead of the peer-group SD.
Comparing with the all-methods group when a method-specific peer group exists.
Ignoring the sign; it shows whether you read high or low.

Keep in mind: These guidelines are general. Follow your proficiency-testing provider's and accreditation body's acceptance rules, and investigate according to your laboratory's quality procedures. Results do not replace clinical judgment.

Frequently asked questions

What is a good SDI?

Values within ±1.0 are generally acceptable; closer to 0 is better.

How is SDI different from CVI?

SDI measures bias against the group mean; the coefficient of variation index compares your imprecision with the group's.

Why use the group SD rather than a fixed limit?

It scales the difference to how much laboratories typically vary for that test.

Sources: Siemens Healthineers Academy, PEPFAR QC workshop job aid "Interpreting PT reports" (HILS1746, 2016)

Checked 9 Nov 2026. Siemens Healthineers Academy, PEPFAR QC workshop job aid “Interpreting PT reports” (HILS1746, effective 6 Dec 2016): SDI = (your result − expected result)/group SD; 0 = no bias, 1.0 acceptable, 1.0–1.9 investigate bias, 2.0 or greater poorest performers; for 5 samples: ≥ 2 SDIs over 1.0 → investigate, mean SDI over 1.5 → significant bias, single SDI over 3.0 → random error, range over 4 SDI → random error possible.

Your numbers