The question is usually asked backwards. People ask which collection device is best, when the answer is entirely determined by two things they have not settled yet: how much sample the assay needs, and who is holding the device.
Settle those two and the field narrows to one or two options. Here is how the choice actually works.
What venipuncture buys you, and what it costs
A venous draw gives you several milliliters of blood, the matrix every clinical assay was validated on, and a trained person doing the collection. If your assay needs volume, or needs serum or plasma at scale, or has never been validated on anything but venous blood, this is not really a choice.
The cost is not the tube. It is the friction. A venous draw requires a phlebotomist, which means either a patient service center the person has to drive to or a mobile phlebotomy visit they have to schedule. For a direct to consumer program, that step is where the funnel breaks. People who bought the test do not complete it. The kit is not the problem. The appointment is.
If your program is clinical, site based, or reimbursed, that friction is normal and venipuncture is usually right. If your program depends on a consumer completing a step alone at home, the friction is the whole business problem.
The capillary options, by mechanism
Every self collection blood device is a variation on getting a small volume out of the capillary bed and stabilizing it. They differ in how they get it, how much they get, and what state it arrives in.
Lancet into an open microtube
The oldest and cheapest approach. A lancet, a fingerstick, and the user milking drops into a small tube. BD Microtainer tubes are the common example here, and most laboratories already handle them without any change to accessioning.
It works, and it produces the widest spread of results of anything on this list. Volume varies by user. Milking the finger causes hemolysis and tissue fluid contamination. Cold hands produce nothing. Expect a meaningful rejected sample rate and design your support process for it.
Lancet with a volumetric collector
Same fingerstick, but the blood enters a device that meters a fixed volume by capillary action. The user cannot over or under fill, which removes the largest source of variability.
Better precision, modestly higher unit cost, and the same fundamental limits: you still need adequate blood flow from a fingerstick, and the total volume is small.
Volumetric absorptive microsampling
An absorbent tip takes up a fixed volume and dries. Nothing liquid ships. No leak, no spill, no cold chain, and a simpler packaging and transport problem.
The trade is on the laboratory side. Analyte has to be eluted off the tip, recovery has to be characterized, and the laboratory has to validate the method for the specific analyte. Not every assay tolerates it.
Push button and vacuum assisted upper arm devices
A device applied to the upper arm or abdomen that uses a small lancet array plus vacuum to draw liquid blood without the user doing anything but press a button.
These solve the two biggest problems with fingersticks at once: they are close to painless, which drives completion, and they collect more volume more consistently than a user squeezing a finger. They cost meaningfully more per unit and they are bulkier, which affects the box. RedDrop is one device in this category.
Dried blood spot cards
Blood applied to a card, dried, shipped ambient. The oldest self collection format and still the cheapest.
The known issues are the hematocrit effect on spot spread, punch to punch variability, and the fact that spot quality depends on the user applying a proper drop rather than layering. Well characterized for the assays it is used for, and a poor fit for anything it has not been characterized for.
The questions that actually decide it
How much sample does the assay need?
Ask for the required volume at the minimum, including dead volume and any repeat testing allowance. This single number eliminates most of the list immediately. A panel needing a milliliter of plasma is not going to run off a fingerstick, no matter how good the device is.
Has the laboratory validated capillary as a matrix?
This is the question people skip. Capillary blood is not venous blood. Depending on the analyte, values can differ, and reference ranges built on venous samples may not transfer.
The laboratory has to validate the method on the specific matrix, from the specific device, at the specific volume. If they have not, the device choice is premature.
How sensitive is the assay to hemolysis?
Fingerstick collection produces more hemolysis than a venous draw. For some analytes that is irrelevant. For others it invalidates the sample. Ask before you choose, not after the first rejected batch.
What completion rate can you live with?
This is a business question dressed as a device question. A device that costs several times more per unit but converts a much higher share of shipped kits into returned samples is usually cheaper per result. Model it per result, not per kit.
Nobody can tell you that number in advance. You find it by running real users, which is what a pilot is for.
Does it ship liquid or dry?
Dry formats simplify the entire outbound and return path. Liquid formats bring closure integrity, orientation, leak risk, and in some cases temperature control. That difference shows up in packaging cost on every unit, forever.
Do not skip the regulatory question
Collection device clearance and intended use vary by device and by claim, and the status of any given device is something to verify directly with the manufacturer and with your own regulatory lead. Do not assume a device that is appropriate for one program transfers to yours, and do not take a marketing page as a regulatory determination.
The device is one component
Whichever mechanism you choose, the thing a user actually receives is a box with a device, instructions, a stabilizer or a card, a return mailer, and a label. Completion rate is driven at least as much by whether that assembly makes sense in someone’s hands as by the device itself.
The best device in an unclear kit underperforms a modest device in a clear one. Test the kit, not the device.
Related reading: how to choose a preservation medium covers what stabilizes the sample once it is collected, and building a kit around a novel collection device covers what happens when nothing off the shelf fits.
