Consumer Genetic Health Reports: What DNA Kits Can Tell You in 2026, and Who Owns the Data

The envelope arrives, the tube gets filled, and six weeks later a phone pings with a pie chart of ancestry and a folder labeled Health. For most of a decade that was the whole story. Then, in 2025, the largest consumer DNA company passed through bankruptcy court and into the hands of a nonprofit research institute, and millions of people suddenly wanted to know a second thing: not just what the spit tube said, but who could read it.

As of early 2026, searches for 23andme health results and for how to delete that data are running side by side, which is a telling pairing. People still want the reports. They are also, finally, reading the consent screens.

This piece does both jobs. It explains what a consumer genetic report can genuinely tell you about disease risk, carrier status and medicine response, how strong the evidence behind each result is, and what the law does and does not do once your genome lives on a server.

What changed recently: why 23andMe health data is trending again

Three dated events explain the current attention. In October 2023 the company disclosed that attackers had used recycled passwords from other sites to enter customer accounts and scrape profile information, eventually affecting roughly 6.9 million users through the relative-matching feature. In March 2025 the company filed for Chapter 11 bankruptcy protection, and several state attorneys general publicly reminded residents that they had a legal right to delete their genetic data and request destruction of stored saliva samples. By July 2025 the sale had closed to a nonprofit research institute founded by the company’s co-founder, with court-approved commitments to keep the existing privacy policy in place.

None of that changed the biology. What changed was the public’s understanding that a genome is a durable asset, and that ownership of the company holding it can move. A consumer genetic test is not covered by the Health Insurance Portability and Accountability Act, known as HIPAA, the federal law that governs how hospitals and insurers handle medical records. MedlinePlus Genetics makes this point plainly: direct-to-consumer companies set their own privacy terms, and those terms can change.

On the medical side, the regulatory picture is older and more stable. The Food and Drug Administration authorized the first consumer Genetic Health Risk reports in 2017 for ten conditions, added a limited BRCA1 and BRCA2 report in 2018, and in the same year permitted pharmacogenetic reports with the condition that results not be used to change treatment without clinical confirmation. Those boundaries still define what a 23andMe health report is allowed to say in 2026, and they are the right place to start.

What a 23andme health report actually reads in your DNA

Start with the technology, because it sets every limit that follows. Consumer kits use genotyping, a method that checks a pre-selected list of DNA positions rather than reading the whole sequence. A typical chip examines several hundred thousand positions. The human genome contains about 3 billion base pairs, the chemical letters of DNA, so a chip samples well under one tenth of one percent of the code.

Each position checked is usually a single nucleotide polymorphism, or SNP, a spot where one letter commonly differs between people. The chip asks a yes-or-no question at each spot: do you carry the version linked to a trait or risk? It does not read the surrounding text, so it cannot find a new or rare mutation that was never placed on the list. This is the core difference from clinical sequencing, which reads a gene letter by letter and can detect variants nobody anticipated.

Everyday comparison: a chip is a spell-checker with a fixed dictionary of known misspellings. Sequencing is a proofreader reading every word. The spell-checker is fast and inexpensive to run, and for common variants it is highly accurate. The FDA review found consumer chips agreed with laboratory sequencing more than 99 percent of the time for the specific variants they test. The weakness is not accuracy on what is measured; it is silence on what is not.

A health report therefore bundles three kinds of output. Health predisposition reports flag single variants with a known effect on disease risk. Carrier status reports look for variants that matter mostly for children. Wellness and trait reports pool many small-effect SNPs into estimates about caffeine metabolism or muscle composition. The evidence behind these tiers differs enormously, which is why each deserves its own look.

What diseases does DNA testing detect at home?

The honest answer is: a short, specific list, and only particular variants within it. The FDA-authorized health predisposition reports cover conditions where a limited number of well-studied variants carry a meaningful, reproducible change in risk. The core group includes late-onset Alzheimer’s disease (the APOE gene), Parkinson’s disease (specific LRRK2 and GBA variants), hereditary hemochromatosis (an iron-overload condition, HFE gene), celiac disease (HLA markers), alpha-1 antitrypsin deficiency (a lung and liver condition), G6PD deficiency (an enzyme condition affecting red blood cells), hereditary thrombophilia (an inherited tendency to clot, factor V Leiden and prothrombin variants), familial hypercholesterolemia (inherited high LDL cholesterol, selected variants) and a three-variant BRCA1 and BRCA2 report.

Notice the pattern. These are mostly conditions where a single gene has a strong, documented effect, where the variant is common enough in at least one population to be worth placing on a chip, and where knowing your status plausibly changes what a clinician would do. Hemochromatosis is a good example: two copies of the C282Y variant explain most cases of European-ancestry iron overload, and iron studies are a simple follow-up.

What the list does not include tells you as much. There is no report that diagnoses type 1 diabetes, lupus, schizophrenia, autism or most cancers, because those conditions arise from hundreds of small genetic effects plus environment. Some companies offer polygenic estimates for type 2 diabetes or coronary artery disease, which pool many SNPs into a percentile. The Mayo Clinic’s guidance on genetic testing is worth repeating here: a predictive result describes a probability, not a future. A detected variant raises risk; it does not confirm that disease will occur, and an undetected variant does not rule it out.

Is there an at-home genetic test for Alzheimer's disease?

Yes, and it is the report people open with the most anxiety. The consumer Alzheimer’s report genotypes the APOE gene, which comes in three common versions: e2, e3 and e4. The e4 version is the strongest common genetic risk factor for late-onset Alzheimer’s disease, the form that begins after age 65 and accounts for the large majority of cases.

The numbers, drawn from the National Institute on Aging’s genetics fact sheet, are these. About one in four people carries one copy of e4, and roughly 2 to 3 percent carry two. One copy raises lifetime risk by about two to three times compared with the common e3/e3 pattern; two copies raise it substantially more, and research in 2024 proposed that e4 homozygosity behaves almost like a distinct genetic form of the disease. Even so, many people with two copies never develop dementia, and a large share of people who do develop Alzheimer’s carry no e4 at all.

That gap between risk and destiny is the entire clinical issue. APOE is a susceptibility gene, not a deterministic one. The deterministic genes, APP, PSEN1 and PSEN2, cause rare early-onset familial Alzheimer’s disease and are not part of consumer chips.

Should you look? Observational studies of people who learned their APOE status with counseling found that most coped well, with short-lived distress that faded over months. Still, expert guidance from the NIA and professional bodies is that APOE testing is not currently recommended as a routine clinical tool, because the result does not yet change prevention advice for most people. Blood-pressure control, physical activity, hearing care and treatment of diabetes lower dementia risk regardless of genotype. If you choose to view the report, do it with a plan for what you will do with either answer.

The BRCA report: what it finds, and the many variants it skips

The consumer BRCA report is the most frequently misunderstood item in any 23andMe health folder. It checks exactly three variants: two in BRCA1 (185delAG and 5382insC) and one in BRCA2 (6174delT). All three are founder mutations most common in people of Ashkenazi Jewish ancestry, where about 1 in 40 individuals carries one of them. In the general population these three account for only a small minority of harmful BRCA variants; more than a thousand different pathogenic variants have been documented across the two genes.

The consequence is asymmetric. A positive result is clinically meaningful and should be confirmed by a certified laboratory, after which a clinician can discuss the established risk figures: a harmful BRCA1 or BRCA2 variant raises lifetime breast cancer risk to roughly 45 to 72 percent and ovarian cancer risk to roughly 11 to 44 percent, compared with about 13 percent and 1 to 2 percent in the general population. A negative result, on the other hand, tells you only that you do not carry those three specific changes. It says nothing about the rest of the gene and nothing about other hereditary cancer genes such as PALB2, CHEK2 or the Lynch syndrome genes.

This is why the NHS and US specialty societies route hereditary cancer testing through family-history assessment rather than through a chip. The question clinicians ask first is who in your family had breast, ovarian, prostate or pancreatic cancer, at what age, and on which side. A strong pattern qualifies a person for comprehensive clinical sequencing regardless of what a consumer report shows. Screening recommendations for people at average risk, such as mammography beginning at age 40 in the United States, are set by national bodies and are not altered by a negative three-variant test.

Carrier status: what a detected variant means for family planning

Carrier reports answer a different question from risk reports. A carrier has one working copy and one non-working copy of a gene for a recessive condition, meaning the condition appears only when a child inherits a non-working copy from both parents. Carriers are nearly always healthy themselves. Consumer kits screen for dozens of such conditions, including cystic fibrosis, sickle cell anemia, Tay-Sachs disease, spinal muscular atrophy and several inherited metabolic disorders.

Being a carrier is ordinary. Everyone carries several recessive variants; most never matter because a partner carries different ones. The report becomes actionable when both partners carry variants in the same gene, which gives each pregnancy a one-in-four chance of an affected child, a one-in-two chance of a carrier child and a one-in-four chance of a child with two working copies.

The limitations mirror the rest of the chip. For cystic fibrosis, consumer panels test a few dozen of the more than 2,000 known CFTR variants, weighted toward those common in European ancestry. Detection rates differ by population, so a negative result reduces but does not remove carrier risk, and reduces it less for people of African, Asian or mixed ancestry. Clinical carrier screening, which many obstetric practices now offer before or early in pregnancy, uses expanded sequencing panels designed to narrow that gap.

A practical sequence works well. Treat a consumer carrier result as a prompt. If one partner shows a variant, the other partner is the person who most needs clinical testing. If both show variants in the same gene, a genetic counselor, a clinician trained to interpret inherited risk and explain reproductive options, becomes essential. MedlinePlus notes that consumer results are generally not accepted for medical decisions without confirmation, and reproductive decisions are the clearest case.

Can a consumer DNA test tell you which medicines will work?

Pharmacogenetics is the study of how inherited variation changes the way a body processes medicines. It is real science with real clinical use. Variants in the CYP2C19 gene, for example, affect how well the body activates the antiplatelet medicine clopidogrel, and variants in CYP2C9 and VKORC1 influence sensitivity to the anticoagulant warfarin. Hospital laboratories test these genes routinely in certain settings.

Consumer reports cover a subset of these variants. When the FDA authorized them in 2018, it attached an unusual condition: the reports may describe how a variant affects drug metabolism, but they may not be used to make treatment decisions, and any change in therapy requires confirmatory clinical testing and a prescriber’s judgment. The reason is practical. A chip can miss rare variants in the same gene, can misclassify people with unusual gene-copy numbers, and cannot weigh the other things that govern drug response, including kidney function, other medicines, age and the condition being treated.

The evidence here is graded moderate to strong for a handful of gene-drug pairs backed by randomized trials, such as CYP2C19-guided choice of antiplatelet therapy after coronary stenting, and weak for many others where the association is observational and the clinical benefit of testing is unproven. Consumer panels do not distinguish between those tiers on the results page.

What this means for a reader holding a report that says, for instance, intermediate metabolizer: do not stop, start or adjust anything. Bring the report to the prescriber or pharmacist, who can decide whether a clinical pharmacogenetic test is warranted and whether the finding is relevant to a current prescription. For most people on most medicines, it will not change anything. For a few, it is a useful flag worth a short conversation.

What the evidence actually says, graded by strength

Evidence in genetics comes in layers, and a consumer dashboard flattens them. Here is how the layers separate.

Strong: single-gene variants with high penetrance. Penetrance is the proportion of people with a variant who develop the condition. Founder BRCA mutations, HFE C282Y homozygosity, factor V Leiden and the APOE e4 association have been replicated across decades of large cohort studies and family studies. The analytic accuracy of chips for these specific variants exceeds 99 percent in FDA validation data. What remains uncertain is each individual’s outcome, because penetrance is never 100 percent and is lower in people without a family history.

Moderate: pharmacogenetic associations. A few gene-drug pairs are supported by randomized controlled trials; most rest on well-replicated observational data. Expert panels publish guidelines for clinical labs, but the benefit of consumer-initiated testing has not been shown in trials.

Weak to very weak: polygenic and wellness scores. A polygenic risk score adds up hundreds or thousands of small SNP effects into a single estimate. These scores were built largely from European-ancestry datasets and perform noticeably worse in other populations. For common diseases they mostly shift a person’s estimate modestly within the normal range, and no randomized trial has shown that receiving one improves health outcomes. Trait reports about taste, sleep or muscle type are curiosities with wide confidence intervals.

Behavioral outcomes: mixed and modest. Observational and trial data on whether people change behavior after a consumer result are largely disappointing; most do not change diet, exercise or smoking, and most do not experience lasting distress either. Harvard Health has summarized this as reassurance on harm, with little proof of benefit.

The practical reading: trust the single-variant reports as prompts for clinical follow-up, treat the pharmacogenetics as a conversation starter, and enjoy the rest as entertainment.

Consumer genetic reports at a glance

The table below sorts the main report types by what they measure, how strong the underlying evidence is and the sensible next step. It is a summary, not a substitute for a clinician’s reading of your own result.

Report type	What it measures	Evidence strength	Key limitation	Reasonable next step
Health predisposition (single gene)	Specific variants such as APOE e4, HFE C282Y, factor V Leiden	Strong for the association; individual outcome uncertain	Tests only listed variants; misses rare ones	Confirm positives clinically; discuss with a doctor
BRCA1/BRCA2 (3 variants)	Three founder mutations common in Ashkenazi ancestry	Strong for those variants	Negative result excludes almost nothing in most populations	Family-history review; clinical sequencing if indicated
Carrier status	Recessive variants relevant to children	Strong for listed variants	Detection rate varies by ancestry	Partner testing; genetic counseling if both carry
Pharmacogenetics	Variants affecting drug metabolism	Moderate; strong for a few gene-drug pairs	Not for treatment decisions without confirmation	Show the prescriber or pharmacist
Polygenic risk scores	Combined effect of many small variants	Weak; poorer in non-European ancestry	No trial evidence of health benefit	Standard prevention regardless of score
Wellness and traits	Associations with caffeine, sleep, taste	Very weak	Wide uncertainty, small effects	Treat as curiosity

Two things stand out when the reports are lined up this way. The strongest evidence sits in the smallest, least glamorous part of the folder, the single-variant results that mostly return negative for most users. And the most eye-catching outputs, the percentile scores and trait predictions, rest on the thinnest footing. A reader who inverts the dashboard’s visual hierarchy, paying close attention to the quiet rows and lightly to the colorful ones, will come away with an accurate picture.

Who owns the data from a direct-to-consumer genetic test?

Direct-to-consumer genetic testing means a product sold straight to the public, with no clinician ordering it or receiving the result. That business model places the data in a legal space most people assume is more protected than it is.

Start with what does apply. The Genetic Information Nondiscrimination Act of 2008, known as GINA, prohibits health insurers from using genetic information to set eligibility or premiums and prohibits most employers from using it in hiring, firing or promotion. The National Human Genome Research Institute’s summary is clear on the gaps: GINA does not cover life insurance, disability insurance or long-term care insurance, does not apply to employers with fewer than 15 employees, and does not cover members of the military or people receiving care through certain federal programs. A handful of states have passed broader laws; most have not.

Now what does not apply. HIPAA governs covered entities, meaning health plans, clearinghouses and providers. A consumer genetics company is ordinarily none of these, so its handling of your genome is governed by its own privacy policy, by state consumer-privacy statutes and by the Federal Trade Commission’s authority over deceptive practices. The policy you accept is a contract, and contracts can be amended.

Research use is the piece that surprises people. Historically about 80 percent of customers at the largest company opted in to research, which permits de-identified data, meaning data stripped of names and direct identifiers, to be analyzed internally and shared with academic and pharmaceutical partners in aggregate. De-identification is strong protection against casual re-identification but not absolute, because a genome is inherently unique.

Law enforcement is the final question. The major companies state that they require a valid court order and publish transparency reports; investigative genetic genealogy cases have mostly relied on separate public upload databases rather than on the testing companies themselves. Policies differ, and reading them is worth ten minutes.

Genetic test data privacy: steps that actually change your exposure

Genetic test data privacy is mostly a set of toggles, and the toggles do different things. Knowing which lever you are pulling matters more than pulling all of them in a panic.

First, separate the three things a company may hold: the raw genotype file, the processed reports and the physical saliva sample. Deleting an account in most services removes the first two from active systems, though companies state that some information may be retained to satisfy legal, regulatory or audit obligations, and anything already incorporated into completed research cannot be unwound. Sample destruction is a separate request; if you consented to biobanking when you registered, the tube may still exist until you ask otherwise.

Second, download before you delete. The raw file is yours to keep, and some clinical genetics services will accept it as background information even though they will not act on it without confirmation. A downloaded file on a personal device carries its own risk, so store it like a passport scan, not like a photo.

Third, revisit research consent and relative-finding features independently of deletion. Withdrawing research consent stops future use of your data in new studies. Switching off DNA relative matching removes you from the feature through which the 2023 account scraping propagated; that incident exposed shared-match lists and self-reported details, not raw genomes, but it illustrated how one weak password can reach strangers’ profiles.

Fourth, enable two-factor authentication and use a unique password. The 2023 breach was credential stuffing, the reuse of passwords leaked elsewhere, and it is the most preventable category of attack.

Fifth, remember that your decision touches relatives. Half your genome is shared with each parent and child, a quarter with grandparents and half-siblings. Privacy here is a family matter, and a short conversation before uploading is a reasonable courtesy.

How many generations until you are no longer related?

The relative finder is the feature that produces the most surprises, and it works on a simple arithmetic of halving. You share about 50 percent of your DNA with each parent and, on average, 50 percent with a full sibling. Each step outward halves the expected share: roughly 25 percent with a grandparent, aunt, uncle or half-sibling; 12.5 percent with a first cousin; about 3 percent with a second cousin; under 1 percent with a third cousin; and around 0.2 percent with a fourth cousin.

Genealogically, relatedness never reaches zero; you and a tenth cousin descend from the same couple eleven generations back. Genetically, it effectively does. Because DNA is inherited in chunks, not as a fine powder, the chance of sharing any detectable segment drops quickly. Nearly all second cousins share measurable DNA. Roughly 10 percent of third cousins share none at all. By fifth cousins, most pairs share nothing a chip can see, and by seventh or eighth cousins the share is close to zero for nearly everyone, even though the paper trail still connects you.

Two practical consequences follow. A match labeled third to fifth cousin can be anywhere in that range, or occasionally a closer relative through an unexpected pathway, because the amount shared at that distance varies widely. And a missing match does not disprove a relationship beyond about third cousin.

The arithmetic also explains why consumer databases became an investigative tool: a database holding a few percent of a population contains a third cousin or closer for most people in it, which is enough for genealogists to triangulate identity. That is a reason the privacy section above treats relative matching as its own decision rather than a default.

Common myths about consumer DNA health results

Myth: a clear health report means I am not at risk. A clear report means none of the listed variants were detected. For BRCA that is three variants out of more than a thousand. For cystic fibrosis carrier status it is a few dozen out of 2,000. Family history, age, blood pressure and smoking remain the dominant predictors for most common diseases.

Myth: a detected APOE e4 means I will develop Alzheimer’s disease. One copy roughly doubles or triples risk; many carriers reach old age with intact memory, and most people with Alzheimer’s disease have at most one copy. The NIA describes e4 as a risk factor, not a cause.

Myth: the chip is inaccurate, so the results are meaningless. For the variants it tests, analytic accuracy exceeds 99 percent. The genuine weakness is coverage, not precision, and confirmatory clinical testing exists precisely to catch the rare chip error before any decision is made.

Myth: my raw file is a clinical record. Raw genotype files include many positions never validated for medical use, and third-party interpretation tools that scan them produce a high rate of false alarms. Published analyses have found that a large share of supposedly pathogenic variants flagged in raw consumer data could not be confirmed on sequencing.

Myth: deleting my account erases everything. Deletion removes your active data and can trigger sample destruction on request, but data already used in completed research and records kept for legal compliance may persist, and relatives’ uploads still reveal shared segments.

Myth: GINA protects me from every kind of discrimination. It covers health insurance and most employment. Life, disability and long-term care insurers are outside its reach in most states.

When to see a doctor about a consumer genetic result

A consumer report is information, not a diagnosis, and the right response to most results is calm follow-up rather than urgency. There are situations, though, where a clinician should be involved soon.

Make an appointment, and ask about a referral to a genetic counselor, if any of the following apply:

• A health predisposition report shows a variant detected for BRCA1/BRCA2, familial hypercholesterolemia, hereditary hemochromatosis, hereditary thrombophilia or Parkinson’s disease, so that confirmatory clinical testing can be arranged.
• You and a partner both carry a variant in the same gene on a carrier report, especially if a pregnancy is planned or under way.
• Your family history is strong regardless of the report: two or more close relatives with breast, ovarian, prostate, pancreatic or colorectal cancer, any relative diagnosed before 50, a relative with a known hereditary cancer variant, or early heart attack or sudden death in relatives under 55.
• A pharmacogenetic report flags a gene relevant to a medicine you currently take. Do not adjust the medicine; bring the report to the prescriber or pharmacist.
• An APOE result is causing persistent anxiety, low mood or sleep disturbance.

Seek care promptly, not because of the report but because of the symptom, if you notice unexplained weight loss, a new breast or testicular lump, persistent change in bowel habit, blood in stool or urine, a leg that is swollen, warm and painful, or sudden chest pain or breathlessness. These warrant assessment regardless of genotype.

Every decision that follows a genetic result, including whether to sequence, whether to screen earlier and whether to change a prescription, belongs with the treating clinician, who can weigh the variant alongside the rest of your history. A genetic counselor is the specialist best placed to explain what a variant does and does not mean, and most health systems offer access through a referral.

Frequently asked questions

What diseases does 23andMe detect?

It flags specific variants for a limited list: late-onset Alzheimer’s disease (APOE), Parkinson’s disease, hereditary hemochromatosis, celiac disease, alpha-1 antitrypsin deficiency, G6PD deficiency, hereditary thrombophilia, familial hypercholesterolemia and three BRCA1/BRCA2 mutations, plus dozens of carrier conditions. It does not diagnose any disease, and it does not test for most cancers, autoimmune conditions or psychiatric illness.

How many generations until you are no longer related?

On paper, never; in your DNA, by roughly the fifth to seventh cousin level. Shared DNA halves with each generational step, from about 12.5 percent with a first cousin to under 1 percent with a third cousin. Around 10 percent of third cousins share no detectable DNA, and most fifth cousins share none, even though a common ancestor still links them.

Is there an at-home genetic test for Alzheimer's disease?

Yes. Consumer kits report APOE status, the strongest common genetic risk factor for late-onset Alzheimer’s disease. The result describes probability, not a diagnosis: one e4 copy roughly doubles or triples risk, and two copies raise it further, but many carriers never develop dementia. Rare early-onset genes are not included. Expert bodies do not recommend routine APOE testing, so consider counseling before viewing.

How much does a 23andMe genetic test cost, and does the kit type change what you learn?

Pricing changes often and varies by tier and promotion, so check the company’s site directly. What matters more is content: the ancestry-only tier includes no health reports, while health tiers add predisposition, carrier and pharmacogenetic reports built on the same genotyping chip. No consumer tier replaces clinical sequencing, and a positive health result still needs laboratory confirmation.

How accurate are 23andme health results compared with a clinical test?

For the specific variants a chip tests, agreement with laboratory sequencing exceeds 99 percent. The gap lies in coverage: a chip checks listed variants only, while clinical sequencing reads entire genes and finds rare or novel changes. That is why a consumer positive is confirmed clinically and a consumer negative is never treated as ruling a condition out.

Can health or life insurers use my genetic test results?

Health insurers and most employers cannot, under the Genetic Information Nondiscrimination Act of 2008. Life, disability and long-term care insurers fall outside GINA in most states, and applications for those products may ask about known genetic results. A few states have passed broader protections, so the answer depends partly on where you live.

Should I delete my genetic test data, and what does deletion actually remove?

That is a personal choice, but know what each step does. Account deletion removes your active data and reports; sample destruction is a separate request; withdrawing research consent stops future studies but cannot recall data already used in completed research. Download your raw file first if you want it, and switch on two-factor authentication whether you stay or go.

What is direct-to-consumer genetic testing, and how is it different from a doctor-ordered test?

Direct-to-consumer genetic testing is sold straight to the public without a clinician ordering or interpreting it. Doctor-ordered tests are chosen for a specific clinical question, usually use sequencing rather than a chip, are performed in certified laboratories, come with professional interpretation, and are protected under HIPAA. Consumer results are governed by the company’s privacy policy instead.

If my BRCA report is clear, do I still need breast cancer screening?

Yes. A clear consumer BRCA report means three specific variants were not found; it says nothing about the rest of BRCA1 and BRCA2 or other hereditary cancer genes. Screening recommendations for average-risk adults, such as mammography starting at 40 in the United States, are unchanged by it, and a strong family history still warrants clinical genetics referral.

What should I do if a consumer report says a variant was detected?

Do not change any medicine or make health decisions on the report alone. Book an appointment with your doctor, bring the report, and ask about referral to a genetic counselor and confirmatory testing in a certified laboratory. If the variant is confirmed, the clinician will discuss what it means for screening, prevention and relatives who may wish to be tested.

References

• MedlinePlus Genetics: What is direct-to-consumer genetic testing?
• MedlinePlus Genetics: What are the risks and limitations of direct-to-consumer genetic testing?
• National Institute on Aging: Alzheimer's Disease Genetics Fact Sheet

This article is for general information only and is not a substitute for professional medical advice. Please consult a qualified doctor about your individual situation.