Use time to investigate an acquisition anomaly

• Distinguish measured time from event order or reconstructed time.
• Use temporal information to generate testable hypotheses.
• Assess how exclusion changes the numerator and denominator.
• Preserve the original acquisition and the rationale for any reanalysis.

Editorially approved by Juan Manuel Ojeda. Independent scientific review has not been completed. An acquisition is a sequence, not just a final cloud of points. Temporal views can expose changes that are hidden after pooling the whole file. Lawrie and colleagues examined sample-to-sample quality monitoring using bead count rate and selected time/fluorescence views in a defined workflow [1]. Independent work on time-domain cytometry QC likewise shows that acquisition artefacts such as clogging can produce spurious events and fluorescence shifts and can be flagged for review [2]. These studies support investigating temporal information; they do not establish a universal automatic clinical rejection or exclusion rule.

Check whether actual time was recorded. FlowJo's kinetics documentation notes that, when no time parameter was collected, it may derive approximate time from event number while assuming constant flow rate [3]. Such a reconstruction cannot independently prove that the event rate was constant; that constancy is part of its assumption.

CFCM investigation framework: start with the observation, list plausible explanations, then identify the evidence that would distinguish them. A transient change could be relevant to fluidics, sample mixing, a pause or other handling. A time plot alone does not establish which explanation is correct.

An exclusion is an analytical intervention. It can change both the population count and the eligible denominator. Do not hide the original file or move a time boundary until a preferred clinical answer appears. Decisions about repeat acquisition, sample handling or reporting belong to the authorised laboratory procedure.

Measured time: timing information recorded during acquisition. Event order: sequence without necessarily knowing the interval between events. Event rate: events per specified measured time interval. Temporal gate: an explicit selection whose effect on the result must be assessed. Root cause: a supported explanation, not simply the first plausible story.

Synthetic file review. Three recorded 30-second intervals contain 30,000, 90,000 and 30,000 events. Their average rates are 1,000, 3,000 and 1,000 events per second. A fluorescence distribution also changes during the middle interval.

This is a time-associated anomaly. It does not prove a clog, and it does not prove that the middle interval contains only artefacts. Review the acquisition record, sample handling and relevant instrument information. Compare scatter and fluorescence behaviour in multiple populations. Record whether the time parameter itself is valid and whether the intervals include pauses.

Before any proposed exclusion, calculate its effect. The middle interval contains 60% of all recorded events: 90,000 / 150,000. Removing it is a major change to the analysed material, even though it occupies only one third of the recorded duration. Its removal may alter precision, representativeness and the applicable reporting claim.

• Assuming event number is an independent measurement of elapsed time.
• Calling every transient change a confirmed fluidic fault.
• Applying a cleanup gate without recording the excluded fraction.
• Forgetting to recalculate the eligible denominator after a justified exclusion.
• Discarding original data so the investigation cannot be reproduced.

A useful troubleshooting note contains the original file identifier, actual observation, time or event intervals, fraction affected, competing hypotheses, checks performed and authorised conclusion. Where a fresh acquisition is appropriate, compare it without silently replacing the problematic record. Technical improvement and defensible documentation should occur together.

The middle interval contains 9 of 15 candidate rare events. Removing it leaves 6 candidate events among 60,000 total recorded events. The remaining fraction is lower. Does that demonstrate that the original 9 were false events?

1. Does a three-fold rate increase identify a unique cause? No.

2. What fraction of the synthetic file would the proposed exclusion remove? 60% of recorded events.

3. Did the challenge's rare-event fraction fall after exclusion? No; both calculations give 0.01%.

[1] Lawrie D et al. A model for continuous quality control incorporating sample-to-sample assessment of optical alignment, fluorescence sensitivity, and volumetric operation of flow cytometers. Cytometry B. 2010. Original study. The study's workflow does not establish a universal clinical exclusion policy.

[2] Meskas J et al. flowCut: An R package for automated removal of outlier events and flagging of files based on time versus fluorescence analysis. Cytometry A. 2023;103:71–81. Open article. Research/computational QC evidence; not a clinical release rule.

[3] FlowJo version 10 documentation: Kinetics overview and How FlowJo Handles Time. Manufacturer software documentation used only for FlowJo-specific measured/reconstructed-time behaviour; it does not establish a clinical exclusion rule.

All acquisition counts, timings and scenarios are synthetic CFCM exercises. No patient data are used.