Interference

Run this study#

Your question: does an added substance shift results by more than you can accept?

Bring: independently prepared specimen pools, the interfering substance and the solvent (vehicle) for matched controls. Choose analyte levels near decision points.

Suggested starting plan: 2 analyte levels × 2 independent pools per level × (1 control + 3 interferent concentrations) × 5 replicates = 80 results. That is 20 results per pool. A neat base aliquot in five replicates per pool adds 20 results, for 100 total. Make the highest concentration your claimed maximum. The replication you need depends on your allowable effect and the assay's variability.

1. Set your limit before testing. Choose Create study. On Set up, under Acceptance limits, choose Enter your own limit for Interference effect and enter the largest acceptable shift (in the study unit, as a percentage of the control mean, or both). It decides each combination and the study. Also enter the interferent unit under Interference settings.
2. Split each pool. Prepare one vehicle-only control and one test aliquot at each concentration, adding matched volumes so both are diluted equally. Give each separate preparation its own pool ID. The CLSI explanation of paired interference studies describes this design.
3. Measure and record each replicate. Enter pool, analyte level, role, interferent concentration, replicate and result.
4. Import and calculate. On Data, choose Import a file and choose your file (start from Blank template (CSV) or interference.csv). Answer any question the dialog asks, then choose Import into this study. Choose Calculate results and review each pool and concentration, and the vehicle effect if you measured a base. On Report, choose Download PDF.

Before reporting: check that every test has its matched control. An interval wholly inside your limit supports that combination. An interval that crosses the limit needs more data. The calculation minimum is 2 replicates in both aliquots for an interval, and 2 pools at each level and concentration to decide a limit. Plan more than that.

Purpose#

How far does the result move when a substance is added to a specimen? You split a pool into a control aliquot, with only the vehicle, and test aliquots with the interferent at known concentrations. Each test aliquot is compared with the control from the same pool.

When to use it#

Use it for hemolysis, lipemia, icterus, or an added drug or metabolite, as a screen at one or a few concentrations or as a concentration response. Each analyte level, for example low and high, is analyzed separately.

For a high sample affecting the next one, use Carryover. For change during storage, use Stability.

Study setup#

1. In your project choose New study. Enter Analyte and Unit, and select Interference under Sample and material effects.
2. Enter Interferent and Interferent concentration unit.
3. Choose Create study, or Create and import data if you already have results.

On Set up, under Interference settings:

• Design: Screening or Concentration response, which expects two or more concentrations per level.
• Claimed maximum interferent concentration (optional). A claim above the highest tested concentration is flagged.

Under Acceptance limits, Enter your own limit offers Interference effect: the largest acceptable effect in the study unit, as a percentage of each combination's control mean, or both. More statistics adds Relative interference effect (%). See Acceptance limits.

Entering data#

A pool is one prepared specimen split into aliquots. It is the independent unit. For each pool and analyte level, enter:

• one control aliquot: vehicle added, no interferent;
• one test aliquot for each interferent concentration;
• optionally one base aliquot: the neat specimen with nothing added.

Choose Import a file; see Importing and mapping. Correct a censored result, or exclude it with a reason; see Missing and censored values.

Fix these before you can calculate:

• marked rows, such as a missing role, a test with no concentration, or a control or base with one;
• a missing interfering substance or unit;
• two controls for one pool and level, or a repeated replicate label within an aliquot. Enter a separate preparation under its own pool ID.

To add concentrations or replicates later, import only the new readings and leave Replace the current rows unticked. Give each new replicate its own label.

Statistics#

Results are reported by analyte level, for each matched pool and concentration.

Effect            = mean(test replicates) − mean(control replicates)
SE                = sqrt( s_test²/n_test + s_control²/n_control )
df (Welch)        = SE⁴ / [ (s_test²/n_test)²/(n_test − 1) + (s_control²/n_control)²/(n_control − 1) ]
Interval          = Effect ± t(1 − α/2, df) × SE
Relative effect % = 100 × Effect / |control mean|      (interval scaled the same way)
Percent limit     = limit % × |control mean| / 100      (then compared in the study unit)
Either            = larger of the study-unit limit and the percent limit
Both              = smaller of the study-unit limit and the percent limit
Vehicle effect    = mean(control) − mean(base)          (reported separately)

• A positive effect means the interferent pushed results up.
• SDs use n − 1. There is no interval when either aliquot has fewer than 2 replicates, or when both have zero spread.
• The relative interval treats the control mean as fixed. Percent effects need a ratio scale and a clearly positive control mean. Without them there is no percent effect, and a percent limit can't be used. With Either, the study-unit limit is then used alone; with Both, the combination can exceed the limit but can't be within it.
• A large vehicle effect means the vehicle itself moves the result. Fix the preparation.

Decisions. Each matched combination is compared with the acceptance limit on unrounded values, and its last column, Result, shows the outcome:

• Not met when the effect is outside the limits.
• Met when the whole interval is within the limits.
• Undecided when:
  • the effect is inside but the interval reaches beyond the limits;
  • there is no interval;
  • fewer than 2 pools were tested at that level and concentration;
  • a test has no matched control.

Sources: NIST/SEMATECH e-Handbook §1.3.5.3 (Welch–Satterthwaite) and §1.3.5.6. See Methods and sources.

Worked example#

interference.csv is synthetic glucose data (mg/dL) with hemoglobin (g/L) as interferent:

• Level low: pools P1 and P2, each with base, control and tests at 1, 2 and 5 g/L; 3 replicates each (30 rows).
• Level high: pool P1 with base, control and tests at 1, 2 and 5 g/L; pool P2 has only a test at 5 g/L and no control (18 rows).

Setup: interferent Hemoglobin, unit g/L, Concentration response, claimed maximum 10, 95% confidence. One acceptance limit: Interference effect within ±3 mg/dL, inclusive, all levels.

What the saved result shows#

Level low, Combinations (mg/dL):

Level high: pool P1 at 1, 2 and 5 g/L gave effects of +1.03, +0.83 and +1.60 mg/dL with 95% CIs −2.01 to 4.08, −2.22 to 3.89 and −1.67 to 4.87. All three are Undecided, because only 1 pool was tested at each concentration and each interval crosses ±3 mg/dL. Pool P2 at 5 g/L has no control, so no effect is calculated. It is listed under Unmatched test aliquots and is Undecided.

Vehicle effect (control − base): P1 low −1.93 mg/dL (95% CI −2.81 to −1.06), P2 low −1.83 (−2.69 to −0.98), P1 high −3.77 (−6.70 to −0.84).

Study status: Criteria not met. A note flags the claimed 10 g/L as above the highest tested concentration, 5 g/L.

With a 3% limit instead, pool P1 at level low (control mean 80.10 mg/dL) was compared with ±2.40 mg/dL, and at level high (control mean 399.43 mg/dL) with ±11.98 mg/dL. With 3 mg/dL and 3% met within either, the larger applies: ±3 mg/dL at level low and ±11.98 mg/dL at level high. Met within both, the smaller applies: ±2.40 and ±3 mg/dL.

Reading the results#

Analyte level switches levels. Effect by interferent concentration plots each combination (◆) at its concentration with its interval. Dots are single test replicates minus the control mean. A limit in the study unit shows as two lines at ±limit. A percent, either or both limit shows as short marks at each combination. Look for effects that grow with concentration and intervals that reach the limit.

Estimates gives the counts and the largest effect. Combinations lists each pool and concentration with its means, effect, interval and outcome. Vehicle effect and Unmatched test aliquots follow when they apply, then Preparation and limit record.

What to do:

• Not met or exceeds limit: recheck concentration, vehicle and pool IDs. If it stands, the interference is real: lower the claim or document the bias.
• Interval crosses the limit: add replicates to narrow the interval.
• No interval: give each aliquot at least 2 replicates that differ.
• Undecided with one pool: prepare a second pool.
• Unmatched test aliquot: add its control.
• Claim flagged: test the claimed concentration or lower the claim.

The study status comes from the acceptance limits. The Why panel under the verdict lists each limit with its outcome. See Criterion outcomes.