Sort every test into one of two piles.

A positive stool test sends you to colonoscopy, where polyps get removed. The test found something and the follow up fixed it. Clean loop. A BRCA1 pathogenic variant changes your surveillance for the rest of your life and may change surgical decisions. A positive HPV test sends you to colposcopy, where precancer gets treated before it becomes cancer.

A variant of uncertain significance on a large gene panel. Nobody knows what it means. Guidelines say explicitly that it must not change your care. It will change your sleep. A moderate penetrance variant found in someone with no family history, where the published risk numbers come from high risk clinic populations and yours is probably lower by an amount nobody can quantify. A borderline prostate number with no prior discussion of what you would do about it.

Pile A is worth having. Pile B is a subscription to background anxiety with no cancellation link. The uncomfortable part is that you do not get to choose which pile you receive. You order the panel. The panel decides.

Before you spend a genome, spend an hour on the phone.

Write down, for parents, siblings, grandparents, aunts and uncles: who had cancer, which cancer, and at what age. Age matters enormously. Breast cancer at 41 is a different signal than breast cancer at 78. Two or more relatives on the same side with related cancers (breast, ovarian, pancreatic, prostate, or colorectal, uterine, gastric) is the pattern that matters.

People with a meaningful family history are far and away the highest yield recipients of this data, and they are the group for whom the panel is most likely to be worth doing. Pathogenic findings concentrate there. Absolute risk estimates are more trustworthy there, because the numbers in the literature were largely derived from families that look like theirs. Management guidance is clearest there. And the result is more likely to change something concrete: an earlier start date, an added MRI, a real conversation about risk reducing surgery, cascade testing for relatives who can then act on it.

If that describes you, get the testing. Get it ordered by a genetic counselor who will spend forty minutes on what the results mean before you give the sample, rather than through a benefits portal that hands you a PDF.

Being the right candidate does not buy you a clean answer. It buys you better odds of one.

A strong family history and a negative panel is a genuinely confusing result. It does not mean the risk was imagined. It means the cause has not been found yet, your relatives still had those cancers, and your screening should probably still be intensified on history alone. People routinely hear "negative" and relax, which is the wrong lesson.

A variant of uncertain significance in a family full of cancer is worse. The natural reading is that the mystery has been solved, and the correct reading is that nothing has changed. Guidelines are unambiguous that such a variant must not drive management, and yet documented harms include people pursuing risk reducing surgery on exactly this basis.

A moderate penetrance finding sits in the middle. The gene is real, the association is real, and the guidance is often thin. Some of these genes carry explicit statements that the evidence is insufficient to recommend added imaging or prophylactic surgery, which is an honest position and a deeply unsatisfying one to receive in the mail.

And a result is never only yours. It tells you something about your siblings, your children, and your parents, none of whom were asked. That conversation is part of the cascade too.

If the history is blank, know what you are buying. In unselected people, roughly 3 to 4 percent get a genuinely actionable finding. Rates of variants of uncertain significance scale with panel size, running from under ten percent on small focused panels to more than forty percent across large panels in heterogeneous populations. The bigger the panel, the more uncertainty. You are far more likely to receive ambiguity than an answer.

The pitch is that the middleman is the problem. No appointment, no waiting room, no speculum, no awkward conversation. The kit arrives at your house.

Sometimes the middleman was the product.

The pre test conversation. Prostate screening carries a grade C recommendation, which in plain terms means this is close enough that it depends on your values, so talk it through first. A kit in the mail has no opinion about your values. It returns a number and hands you the consequences.

The rest of the exam. A photo of the mole you noticed is not a full body skin check. The dangerous lesion is often the one you did not photograph, because you could not see it.

The reflex test. A self collected vaginal sample cannot be run for cytology. A positive one sends you into the clinic anyway, for the exam you were trying to skip.

The chart. Results land in the vendor's system, not your doctor's. Nobody reconciles them against what you already had done. You become the integration layer.

Your genome ends up with a company whose business model is younger than your mortgage. Ask the boring questions. Where is it stored, for how long, is it used for research, can you delete it, what happens if the company is acquired or folds, what happens when your employer changes vendors next year.

Then the one people miss. The 2008 Genetic Information Nondiscrimination Act protects you from genetic discrimination in health insurance and employment. It does not cover life, disability, or long term care insurance, and an equivocal or positive result can change your insurability in all three. Those underwriters can ask, and you have to answer. If you have not bought that coverage yet, the sequencing matters. Buy first, test second. (NHGRI on genetic discrimination.)

This is the only irreversible item in the box. A stool test you can repeat next year. A genome you cannot un-know.

The intuition to internalize: when a disease is rare, most positive results are wrong. This is not a flaw in the tests. It is arithmetic, and it is unavoidable.

Specificity is about 94 percent. Screen 1,000 average risk adults and roughly 60 get a positive result, of whom about 2 to 4 have cancer. Some others have precancerous polyps worth removing. So most people sent for colonoscopy do not have cancer.

That is fine, and here is why. The confirmatory test is also the treatment. You go, they look, they remove polyps if present, you are done for years. The false positive costs you a day and a prep, not an organ.

At the common 4.0 ng/mL threshold, sensitivity for any prostate cancer is about 20 percent. Read that again. Four out of five cancers are missed at that cutoff, which means a normal result is much weaker reassurance than people assume. Meanwhile, among the cancers screening does find, a third to a half would never have caused symptoms in a lifetime. The trials show a modest reduction in prostate cancer deaths and no reduction in death from any cause.

Both failure modes at once. It misses a lot, and much of what it catches did not need catching. Which is why every guideline insists on a conversation first, and why a kit that skips the conversation is the weakest item in the box.

It looks like a finding. Published rates run from under ten percent on small focused panels to more than forty percent on large ones in heterogeneous groups. The bigger the panel, the more uncertainty, and the more likely you are to end up with a result that leaves you anxious and your doctor with no next step.

Those rates also run higher in Black, Asian, and Hispanic patients, for an unglamorous reason: people of European ancestry are the most common reference genomes in the comparison databases. The test is not worse. The library it is checked against is thinner. Of the variants eventually reclassified, 80 to 90 percent turn out to be benign, sometimes years later.