Carryover

Run this study#

Your question: does a high sample raise the result of a low sample measured after it?

Bring: one very low material and one very high material, with enough volume for 11 low and 10 high measurements: 21 results per run. Choose concentrations that challenge the analyzer in its intended use. Patient material is best, but suitable control or EQA material also works. This follows the pSMILE collection design.

1. Set the limit first. Choose Create study. On Set up, under Carryover · 21-position sequence, choose the built-in 3 × SD of low-after-low results, or enter an absolute error limit. The SD rule judges the effect against the run's own scatter. If you have a limit set from clinical need, enter it as an absolute limit instead.
2. Prepare and label the positions. Use Start a blank 21-position run on Data or the blank collection template. Use the same low and high material throughout.
3. Measure in exactly this order, with no other samples or tests in between:

Position:  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Role:      L L L H H L H H L L  L  L  H  H  L  H  H  L  H  H  L

4. Record all 21 results with run ID, position, role, material ID and result. Type them into the blank run, or fill in the template and on Data choose Import a file and choose your file. Answer any question the dialog asks, then choose Import into this study. If a run is interrupted, repeat it in full under a new run ID.
5. Calculate. Choose Calculate results. Review each run under Results, then on Report choose Download PDF.

You get: for each run, the rise in low results after a high, the error limit and a decision. Carryover must be below the limit. A carryover equal to the limit fails. One complete run is enough to calculate. The study status is Criteria met when every run meets its limit.

Purpose#

The study compares low results that follow a high with low results that follow a low. It uses the fixed 21-position sequence.

When to use it#

Use it for a sample-to-sample carryover check on one analyzer. For an added substance such as hemoglobin, use Interference. For storage time, use Stability. For the low material's precision alone, use Repeatability.

Study setup#

1. In your project choose New study. Enter Analyte and Unit, select Carryover under Sample and material effects, and choose Create study, or Create and import data if you already have results.
2. On Set up, under Carryover · 21-position sequence, choose the Error-limit basis: 3 × SD of low-after-low results (the default) or User-defined, in your unit. A user-defined limit needs an Error limit.
3. Enter your runs (see Entering data).
4. Choose Calculate results.

The error limit is the only acceptance rule.

Entering data#

Start a blank 21-position run or Append a blank 21-position run adds 21 rows with the next run ID, positions, roles and the material IDs Low and High. Type the results. Or fill in the blank template and choose Import a file. See Importing a file.

The experimental unit is the run.

The calculation stops if a row is marked, for example a blank result or two rows in a run sharing a position.

A run is calculated but Undecided when:

• a position is missing, excluded or outside 1–21
• a role doesn't match the sequence
• the low and high material IDs conflict within the run
• the run's results span more than one day
• the limit is 3 × SD and all five LL results are identical, so the SD is zero
• the high-material mean is not above the low-material mean
• a user-defined limit has no value

Statistics#

For each complete run:

LL mean      = mean of results at positions 2, 3, 10, 11, 12
HL mean      = mean of results at positions 6, 9, 15, 18, 21
LL SD        = sqrt( sum((LL result − LL mean)²) / (5 − 1) )
Carryover    = HL mean − LL mean          (signed, study unit)
Default limit = 3 × LL SD
Decision     = meets the limit if Carryover < limit

• Decisions use full-precision values. A carryover and limit that look equal on screen can still differ, in either direction; a note says which side the exact carryover is on. With the SD limit, the test is Carryover² < 9 × LL variance for a positive carryover.
• A negative carryover (lows after highs read lower on average) is below any positive limit and passes. A note points it out, because the check looks for an increase only.
• Sources: pSMILE Carryover Guidelines VAL 2015 v1.1, reviewed 2025 (strict less-than rule with 3 × LL SD). See Methods and sources.

Each run is judged by one limit, Carryover below error limit, shown as Met, Not met or Undecided in the Why panel. Any run that is not met gives Criteria not met. Otherwise, any undecided run gives Undecided. When the LL SD is zero, a user-defined limit can still decide. See Limit outcomes.

Worked example#

Download carryover.csv: one synthetic run R1, creatine kinase in U/L, low material LOW-01 (18 to 24 U/L) and high material HIGH-01 (4988 to 5021 U/L). Setup: 3 × SD of low-after-low results.

LL results (positions 2, 3, 10, 11, 12) are 19, 19, 20, 21, 21 U/L. HL results (6, 9, 15, 18, 21) are 20, 21, 22, 23, 24 U/L.

Reading the results#

Choose a run with the Run selector.

Results in measurement order plots the high material on top and the low below, on positions 1–21. LL points are circles, HL points diamonds and the initial low a square. Lines mark the LL and HL means. Look for HL diamonds above the LL line, or a trend along the run.

Carryover effect and limit gives LL and HL n, mean and SD, the carryover, the error limit and its basis. The outcome and, for an undecided run, its reason are in the Why panel under the verdict and in the notes above the results.

Run … · source results lists all 21 rows in position order, including excluded rows with their reason.

Look into any outcome other than Met, HL results rising along the run, or an LL SD so large that the limit is wider than you would accept.

If a run is undecided, read the reason. A missing or excluded position, or a run measured in another order, means repeating the run. A run that fails on good data shows carryover from the high material into the next low.

Common mistakes:

• Using one material ID for both materials.
• Reusing a run ID for two experiments.