What Does the Combined Paternity Index Mean?

The Combined Paternity Index (CPI) is the number on a DNA paternity report that measures how strongly the DNA points toward the tested man being the father.1US Diagnostics Center. Combined Paternity Index (CPI) Explained It is not a percentage chance. It is a comparison between the man you tested and a random man from the population.2National Institute of Justice. Population Genetics and Statistics for Forensic Analysts | Parentage and Relatedness

How to Read the CPI Number on Your Report

Every CPI weighs two explanations for the same DNA results. In one, the tested man is the biological father. In the other, an unrelated man from the same general population is.1US Diagnostics Center. Combined Paternity Index (CPI) Explained The number says how many times better the first explanation fits the DNA than the second.3PubMed Central. Early noninvasive prenatal paternity testing by targeted fetal DNA analysis So a CPI of 1,000 reads as “the results are 1,000 times more likely if he is the father than if a random unrelated man is.”

Where the number falls tells you which way the evidence leans:

On a real report, you’ll usually find a table of markers, each with its own score, and the CPI at the bottom. Next to it there is typically a paternity percentage. A report that says “excluded” usually shows a CPI of zero or no CPI at all. If the result goes to court, the CPI is the figure an expert will usually point to as the strength of the evidence.

A CPI of 10 is modest support, and a CPI in the millions is overwhelming. Values near 1 mean very little. A CPI of 1.5 is barely evidence, because the DNA fits both explanations almost equally well. That usually reflects too little informative DNA, so a lab would normally test more markers instead of drawing a conclusion. A very small CPI such as 0.001 means the DNA fits an unrelated man far better than it fits him, which points away from paternity.

A common misreading is to treat the CPI as the chance he is the father. A CPI of 10,000 does not mean a 1-in-10,000 chance he isn’t. It compares him only with an unrelated random man, not with his brother or father.

How Labs Arrive at the CPI

Labs test many separate spots in the DNA, called markers, and give each one its own score, the paternity index (PI).4What Does the Combined Paternity Index Mean?. What Does the Combined Paternity Index Mean? At each marker the child carries one piece from the mother and one that must have come from the father.2National Institute of Justice. Population Genetics and Statistics for Forensic Analysts | Parentage and Relatedness The lab asks how likely the tested man was to pass on that piece compared with a random man.

Rarity drives the answer. The lab looks up how common each piece of DNA is in a reference population. If many men share that piece, a match says little. If few men have it, a match strongly points toward the tested man, so a rare piece earns a bigger score than a common one. The reference population matters too, because scores depend on how common each DNA piece is in the comparison group.5PubMed Central. Evaluating Methods to Correct for Population Stratification when Estimating Paternity Indexes

Here is one marker in action. Suppose the child has a piece the mother does not have, so it must have come from the father, and the tested man carries it. The lab compares the chance this man passes it on with the chance a random man does. If that piece makes up only about 1 in 10 copies of the marker in the population, the score is about 5 or more. If the man doesn’t carry the needed piece at all, that marker’s score drops sharply, and the lab then considers whether a mutation could explain it.

The individual marker scores are then multiplied together to give the CPI.2National Institute of Justice. Population Genetics and Statistics for Forensic Analysts | Parentage and Relatedness Markers are chosen so that they are inherited independently of one another, which is what makes multiplying fair instead of adding. Twenty markers each scoring around 2 to 5 can multiply into a CPI in the millions. Because it’s multiplication, one very informative marker can lift the CPI a lot, while a weak one changes it only a little. That’s why adding more markers is the usual way to strengthen a borderline result.

CPI Versus the Probability of Paternity

The CPI is an odds figure, not a probability.2National Institute of Justice. Population Genetics and Statistics for Forensic Analysts | Parentage and Relatedness Reports usually also show a probability of paternity, calculated from the CPI with a starting assumption of a 50% chance.6Genetics and Molecular Research. PedExpert: a computer program for the application of Bayesian networks to human paternity testing With that starting point, the probability equals the CPI divided by the CPI plus 1. A CPI of 100 becomes about 99.0%, and a CPI of 10,000 becomes about 99.99%. Once the CPI passes the thousands, the percentage changes only in later decimal places, so a CPI of 10,000 and one of 1,000,000 look almost identical as percentages.

You can check your own report this way. A CPI of 99 gives 99 out of 100, or 99%. A CPI of 19 converts to 95%, so even a modest CPI can produce a percentage that looks high. If your report’s percentage doesn’t match this conversion, the lab probably used a different starting chance, and the report should say so.

The 50-50 assumption means the lab treats the man as equally likely to be the father or not before any testing. It ignores everything else known about the situation, which is deliberate, for fairness. If other evidence makes paternity more or less likely, the real-world chance would shift.

A paternity percentage of 99.99% does not mean the test is 99.99% accurate. It is the chance of paternity under the 50-50 assumption, not an error rate. The remaining 0.01% isn’t a chance the lab made a mistake; it comes from the same odds calculation. Sample mix-ups and handling errors are separate risks, which is why proper collection and an accredited lab matter.

What CPI Counts as Proof of Paternity

The AABB standards that accredit US relationship-testing labs set a minimum CPI of 100 for parentage cases.7AABB. Relationship Testing Technical Report for Testing in 2024 Many labs set their own higher minimums, such as 10,000 to 1 or 100,000 to 1, and half of labs have a policy minimum above the AABB standard.7AABB. Relationship Testing Technical Report for Testing in 2024

A minimum gives labs a consistent line, so they don’t report paternity on evidence too weak to be dependable. It protects men and children from decisions based on thin evidence, which matters because results can affect support, custody and family relationships. Falling below a lab’s minimum is not an exclusion. It only means the evidence wasn’t strong enough for that lab to state paternity.

If your CPI is below the lab’s cutoff, the lab may report the result as inconclusive rather than a match. It often adds more markers or asks for the mother’s sample. Ask what minimum the lab uses and whether retesting is included. Above the lab’s minimum, the report will state that paternity is not excluded or is supported. Below it, expect inconclusive wording.

For a legal test, courts usually require an accredited lab and a chain-of-custody sample collection. Home kits may give a CPI but often aren’t accepted in court. Check that the report lists the CPI, not only a percentage.

When a Relative Might Be the Father

The standard CPI compares the tested man with an unrelated man, so it doesn’t hold up when a relative could be the father. Changing the comparison from a random man to the tested man’s brother can change the result dramatically.8PubMed. Beyond traditional paternity and identification cases. Selecting the most probable pedigree Brothers, fathers and sons share much of their DNA. Brothers share about half on average, so a brother can carry many of the same pieces the child inherited from the father’s side. Each shared piece adds to a standard CPI, because the calculation assumes the alternative is a stranger.

In tests without the mother, some uncles showed no DNA mismatches with the child and reached paternity probabilities above 99.97%.9PubMed. Possible pitfalls in motherless paternity analysis with related putative fathers A high CPI alone can’t separate brothers, so labs must be told about related candidates. A father or son of the tested man shares half his DNA too, so the same concern applies.

Tell the lab up front whether any possible fathers are brothers, father and son, or cousins. Ideally, test all the possible men rather than just one, and ask the lab to report the comparison against the relative, not just a random man.

A relative-adjusted calculation replaces the random man with the specific relative, so the CPI asks whether the tested man fits better than his brother. This usually lowers the CPI substantially, which is the honest result when two men share so much DNA. Testing both men is usually the best way to separate them, since their DNA will differ at some markers.

A different problem arises when the possible father is related to the mother. Special formulas exist for that case, and they give a more cautious score than the standard method.10PubMed. Calculation of the Paternity Index for the Alleged Father Related to the Child’s Mother

Why Tests Without the Mother Are Weaker

A test without the mother’s DNA is analysed differently from a mother-child-father test, but the two are often handled as if they were the same.11PubMed. Motherless case in paternity testing Some cases can be wrongly decided as a result, and labs should use a different standard for motherless tests.11PubMed. Motherless case in paternity testing

The reason is that without the mother, the lab can’t tell which of the child’s two pieces came from her and which from the father, so every marker is less informative. Each marker’s score is smaller, and the total CPI is lower than a full trio would give for the same true father. A weaker score also leaves more room for a related or unrelated man to fit by chance.

A motherless result can still be correct, and a high CPI from a good lab carries real weight. The caution is about how the result is judged: the same CPI means less when the mother is absent, so the lab’s minimum should be stricter. Check whether your report says the mother wasn’t tested, and ask what minimum the lab applies for such cases. For legal disputes, a full trio is the stronger option.

Mutations, Mismatches and Other Pitfalls

DNA sometimes changes slightly when passed from parent to child. So labs build a small mutation chance into the score rather than excluding a man over one mismatch. Some mutations go unnoticed, though, and ignoring them makes the CPI look stronger than it should.12Forensic Science International. What Does the Combined Paternity Index Mean?

On a report, a mutation appears as a marker where the child’s piece doesn’t match what the tested man could have passed on, even though every other marker fits. Labs usually add a small mutation rate to that marker’s score, which lowers the CPI but doesn’t discard the result. That’s how a true father can still receive a high CPI despite one mismatch. Look for a note about a possible mutation at one marker, and if you’re unsure, ask whether extra markers were added.

One mismatch is treated differently from several. A single mismatch is plausible as a mutation, because mutations at individual markers are uncommon but do happen. Several mismatches across independent markers are very unlikely to all be mutations, so they usually point to exclusion. Most labs require two or more mismatches to exclude, but they differ, so ask for your lab’s rule.

Hidden DNA variants are a separate cause. Sometimes a piece of DNA fails to show up on the test, so the person looks like they carry only one piece at that marker, and father and child look mismatched.13SAMJ: South African Medical Journal. STR null alleles complicate parentage testing in South Africa Allowing for this raised the CPI in every such case studied, in one from 7 to over 442,000.13SAMJ: South African Medical Journal. STR null alleles complicate parentage testing in South Africa Labs that allow for it often retest that marker or use another method to confirm.

CPI in Prenatal Paternity Tests

A pregnant woman’s blood carries small fragments of the baby’s DNA, which come from the placenta and generally match the baby’s. Prenatal tests compare the man’s DNA with those fragments and produce a CPI the same way.14PubMed Central. Noninvasive Prenatal Paternity Testing (NIPAT) through Maternal Plasma DNA Sequencing: A Pilot Study It uses the same comparison as any other test: him versus an unrelated man.3PubMed Central. Early noninvasive prenatal paternity testing by targeted fetal DNA analysis

To get at the baby’s DNA, the lab separates the fragments from the mother’s own DNA, using her sample as a reference to tell the two apart. It then checks which pieces the baby has that the mother does not, and compares them with the man’s DNA.

A prenatal CPI can be less certain. The baby’s DNA makes up only a small share of the DNA in the mother’s blood, especially early in pregnancy, so there is less signal to read. Some markers may be unclear, and the CPI may be lower than from a test after birth. If the result is borderline, a repeat test later in pregnancy or after birth can confirm it. Before testing, ask how early in pregnancy the lab tests, what minimum CPI it requires for reporting, and whether the result is accepted for legal use.