NakedSignal
Write to us

Change explorer · Haematology and immunology

Point-of-care CD4 count vs laboratory flow cytometry

Public dataBuilt from a published open-access study.

Flow cytometry, laboratory platform compared with point of care CD4 and haemoglobin analyser: 247 paired results from a published study by Sagnia et al. (2024), re-analysed by NakedSignal as if one laboratory had switched from the first system to the second. At 100, 200, 350, 500 cells/µL, 32 results fall in a different category on the second system than on the first, against about 33 expected from repeat testing alone. Here the count sits below the repeat-testing expectation, but that expectation uses a default repeat-test imprecision, which can overstate repeat-testing noise, and the count cannot be told apart from what repeat testing alone would move, so it is not evidence of fewer than noise. A laboratory’s own measured imprecision decides. The deposit holds 247 pairs; the paper reports 230 valid results and its exclusions cannot be identified, so all 247 are used.

Haematology and immunology

CD4 absolute count, cells/µL

Flow cytometry, laboratory platform compared with Point of care CD4 and haemoglobin analyser. 247 paired results.

Compare it with the other comparisons in the explorer

Comparison line

The new system reads 3.7% higher than the old system on average. The comparison line (Passing-Bablok) has a slope of 1.044 (95% interval 1.014 to 1.071) and an intercept of -0.03.

Public data
Comparison line, CD4 absolute count (cells/µL)Comparison line with slope 1.044 (95% interval 1.014 to 1.071) and intercept -0.03, drawn against the line of no change. Decision limits drawn at 100, 200, 350, 500 cells/µL. Where the comparison line sits above the line of no change, the new system reads higher; below, lower.200400600200400600100200350500No changeComparison lineOld system, cells/µLNew system, cells/µL

Fig. 1. The comparison line against the line of no change, over the range of the decision limits (100, 200, 350, 500 cells/µL). Dots mark where each line meets each limit; the gap between them is the bias at that limit.

Crossing each limit

Results that fall on the other side of a decision limit on the new system than on the old, read as if the new system had replaced the old, against the number repeat testing on one unchanged system would move. The program reports the count across all limits together.

All decision limits together (100, 200, 350, 500 cells/µL)

Crossed between the two systems: 32

Expected from repeat testing alone: about 33

Grade and warnings

  • Re-test band: full (at least the protocol minimum of pairs).
  • Band half-width around each limit: 6.49%.
  • Results re-tested on data the band was not fitted to, median: 15%.
  • Repeat-test imprecision: default value.
  • Note: Fewer than 5 results crossed beyond what repeat testing alone would move; excess caught is not meaningful here.

Source

Sagnia B, Mbakop Ghomsi F, Moudourou S, Gutierez A, Tchadji J, Sosso SM, Ndjolo A, Colizzi V (2024). Accurate and reproducible enumeration of CD4 T cell counts and Hemoglobin levels using a point of care system: Comparison with conventional laboratory based testing systems in a clinical reference laboratory in Cameroon. PLoS One. doi:10.1371/journal.pone.0297790 (opens in a new tab); PMC10954178 (opens in a new tab). Licence: CC BY 4.0 (opens in a new tab). Re-analysed by NakedSignal. The deposit names the systems.

Show the numbers
All values for CD4 absolute count, Flow cytometry, laboratory platform to Point of care CD4 and haemoglobin analyser
FieldValue
Paired results used247
Comparison line slope1.043741
Slope, 95% interval1.014465 to 1.071434
Comparison line intercept-0.030269
Mean difference, new minus old, as % of old3.734
Decision limits (cells/µL)100, 200, 350, 500
Crossed between the two systems32
Expected from repeat testing alone33.04
Band half-width (%)6.49
Re-tested, unseen data, median (%)15.18
Re-tested, unseen data, range (%)12.96 to 17
Excess caught, line and band (%)none
Excess caught, line alone (%)none

Related

Other comparisons from the same study:

Guide: Crossings by chance

Every study behind the explorer, with its licence: public data. How the counts are made: how it works.

Changing a system like this one?

Request a first site report