How a blood sample becomes a number on a report

It is tempting to imagine a laboratory result as a direct reading of the body, the way a thermometer reads a room. The reality involves several handling steps, each of which can shift the number, and understanding them explains a great deal of what otherwise looks like inconsistency.

Laboratory scientists divide the process into three phases: what happens before analysis, the analysis itself, and what happens after. The first phase is where most errors occur, and it is the one that happens outside the laboratory.

Before the needle

Several things about the person are already affecting the result before any sample is taken. Time of day matters for cortisol, iron and testosterone, all of which follow a daily rhythm. Recent food matters enormously for triglycerides and glucose, and modestly for several other markers.

Posture matters more than most people expect. Standing for a period concentrates the larger molecules in blood, so proteins, calcium bound to those proteins, and cholesterol all read slightly higher after standing than after lying down. This is one reason results taken in a hospital bed can differ from results taken in a clinic.

Recent intense exercise raises several muscle-derived markers, including creatine kinase and aspartate aminotransferase, and can nudge creatinine upward. A gym session the previous evening is worth mentioning if a muscle or liver marker comes back unexpectedly.

The draw itself

A tourniquet left in place for more than a minute or so begins to concentrate the sample, raising the measured value of anything bound to protein. It can also cause cells to leak potassium into the surrounding plasma.

The order in which tubes are filled matters, because the additives in each tube are different and a small carry-over between them changes results. Tubes are colour-coded by additive, which is why a single draw fills several of them.

Blood drawn from an arm with a drip running into it can be diluted by the drip fluid, sometimes dramatically. A sodium or glucose result that makes no sense alongside the rest of a panel is occasionally explained this way.

Getting to the analyser

Once drawn, a sample begins to change. Cells continue to consume glucose, so a glucose sample left standing without a preservative reads progressively lower — which is why glucose tubes usually contain an inhibitor.

Haemolysis, the rupture of red blood cells, is the most common sample problem. It releases the contents of those cells into the plasma, raising potassium and several enzymes, and it interferes optically with a number of other tests. Laboratories detect it and often reject the sample rather than report a result they know is wrong.

Temperature and time both matter. Some markers are stable for days, others for hours. A sample collected at a rural clinic and couriered to a central laboratory has a different history from one drawn twenty metres from the analyser.

The measurement

Modern analysers measure most markers indirectly. An antibody binds the target and produces a signal proportional to how much was there; an enzyme reaction consumes the target and the rate of consumption is measured; light scattering, electrical conductivity and mass spectrometry each have their place.

Every method has a range over which it is reliable, and results outside that range are diluted and re-run rather than reported directly. Every method also has substances that interfere with it. Biotin, taken in high doses in hair and nail supplements, is a well-known interferent for several immunoassays.

Laboratories run control samples with known values alongside real ones, throughout the day, to detect drift. Results are not released if the controls fail, which occasionally explains a delay.

After the number exists

The result is compared against stored reference intervals to generate flags, checked against critical limits, and in many laboratories checked against the patient's previous result. A large unexplained change from a recent value triggers a review before release, on the reasonable assumption that people change more slowly than samples get mixed up.

None of this is visible on the report. What arrives is a number, a unit and an interval — the compressed output of a process with many steps, most of which went right.