Paternity Evidence

The general concept of testing for paternity is centered on the establishment of information about hereditary factors that either exclude an individual from consideration of being the biological father of a child, or reveal a convincing pattern of consistency that supports a claim of biological paternity. Exclusion can be absolute—it is indeed possible to disprove a person's role as biological father. It is not possible, however, to make a positive proof of paternity. This side is always a probability calculation. Thus, the development of paternity evidence involves both physical testing, through DNA or other biochemical markers, and probability calculations using the laws of probability.

One important aspect of paternity testing is the development of a list of potential candidates for paternity. Since the timing of conception is fairly tightly clustered around the middle of the menstrual cycle, the mother of a baby generally knows with a fair amount of certainty who the father is, or knows the list of possible candidates with whom she has had sexual intercourse near enough to the time of conception for determining paternity to be a realistic possibility. In some cases, and for various reasons, the mother may not have a conscious awareness of all of the events surrounding the pregnancy. This may be the situation in cases of rape by an unknown assailant, intercourse which has taken place under the influence of drugs or alcohol, or when the mother is mentally retarded or has certain forms of mental illness. The first step in the process of paternity testing is to determine the candidates for whom testing makes sense. This type of testing is centered on elimination or retention of individuals who have been placed on the list of reasonable candidates.

In previous years, testing was focused on the testing of blood group types and the evaluation of biochemical markers for which there were significant differences among individuals in the population. This testing seems crude compared with the more precise and informationally rich DNA marker systems for testing that are currently in use. In principle, any marker that is inherited from the parents can be used as a part of the testing process, however.

In the laboratory testing phase of the analysis, the laboratory chooses a number of markers, which have different forms in the general population, for analysis. These markers are called polymorphic markers, meaning each marker has many forms. Simple markers may have just two forms, and each individual has two copies of each marker. Thus, with these simple two-marker systems, a person would fall into one of three categories: he could have two copies of the first form, one copy of each of the two forms, or two copies of the second form. There are just three possibilities for anyone in the population, and it is common to match with a genotype consistent with paternity just by chance. The greater the number of possible forms, however, the greater the number of different combinations in the population, and the lower the likelihood of matching purely by chance. For example, if there are three different forms of a marker, there are 6 combinations possible; four forms yields 10 different combinations; five forms yields 15 combinations. For many of the genetic markers available for testing, such as short tandem repeats, there may be 10 to 20 different forms and therefore many, many combinations possible.

The testing strategy would then be to select several markers for testing, and to test the mother, the child, and the suspected father for each of the markers selected. Starting with the child, for each marker studied, one would ask which of the two copies that the child has came from the mother. For highly polymorphic systems it is often true that one and only one of the child's markers could have come from the mother. The remaining marker that the child carries must have come from the father. Sometimes, the child and mother match exactly for both forms of the marker, and it is not clear which one came from the mother and which came from the father. In this case, if either form matches one of the forms of the marker that the father carries, it is consistent with paternity.

As an example, let us say that there is a marker we will call X that has twelve forms that can be found in different people. We will let X be a trinucleotide repeat that is found to have anywhere from six to seventeen copies in normal individuals in the population. Each person will have two alleles of X, one that was inherited from his mother and one that was inherited from his father at conception. Upon testing, let us say that the child has one allele that has 7 copies of the repeat, and the second allele has 11 copies of the repeat. In the mother, we find one allele that has 7 copies of the repeat, and the other has 8 copies of the repeat. We know that the mother must have passed along the allele with 7 copies of the repeat. The other allele that the child has must come from his father. We can now exclude any suspected father who does not carry at least one allele of X that has 11 copies of the repeat. But what if the father does have an allele that has 11 copies of the repeat? Does this prove paternity? No, this could be a match purely by chance. While it is consistent with paternity, it is not conclusive by itself.

In real practice, one would not use just a single marker, even if it were highly polymorphic. One would generally include several informative markers to increase the chance that the suspect will be eliminated by failing to match. By choosing markers with a lot of variability in the population, the chance of matching can be minimized. For a person to be retained as a candidate for paternity, matching has to occur for all of the markers. Even a single inconsistency can eliminate a person from consideration.

If a person matches on all of the markers included in testing, and if those markers are reasonably informative markers for testing, the individual being tested is the presumptive father. In this case, it will be necessary to compute the likelihood of a person matching purely by chance using the simple laws of probability.

The probability of an individual carrying a marker of some given size can be found by studying a large number of people and computing the number who carry the marker divided by the total number of people studied. Suppose 500 people are studied, and 50 of those people have are found to have at least one copy of marker X with 11 repeats; the chance is 50/500 or 0.10 of carrying a marker of that size.

One rule of probability is that the probability for both of two different events happening is found by multiplying their individual probabilities together. Likewise, to compute the chance of three or more separate events each happening one would multiply each of their individual probabilities together. When the probabilities are each small, the product of the combined probabilities becomes very small. It is possible to end up with likelihood of paternity that says that the chance of matching purely by chance is one in a million or less.

A reasonable question that many people ask is what is the chance that the testing is wrong. The simple answer is that the chance of being wrong when the father has been excluded by DNA testing is very near to zero. This assumes, however, that the specimen that was studied actually came from person that you think is being testing. Great care must be given to ensuring that the blood sample or other specimen that is taken for paternity testing actually comes from the person that is suspected as being the father. It is standard practice for laboratories that perform paternity testing to document a chain of custody for the specimen from the time it is drawn, until the time it reaches the laboratory for testing.

What about the chance of being wrong when the testing is consistent with paternity? As the result is expressed as a probability statement, it always remains true that there is a possibility of a match by chance. When there is reason to suspect that this is the case, the study of additional markers can further reduce the likelihood of a match by chance. While it is not possible to get this probability to zero, the probability can always be further reduced by adding additional markers. This adds expense, however, and most people will quickly realize that such expense is not warranted unless there is some compelling reason to doubt the findings. It is rarely the case that a person enters paternity testing without some fairly high likelihood that he is in fact the father.

The last twenty years of the twentieth century saw dramatic developments in the understanding of genetics and the development of markers that can be used in paternity and forensic testing. Compared with the testing available in previous generations, determination of paternity is now extremely reliable and relatively inexpensive.

see also DNA; DNA evidence, social issues; DNA typing systems; Evidence, chain of custody; Gene; Genetic code; RFLP (restriction fragment length polymorphism); Statistical interpretation of evidence; STR (short tandem repeat) analysis.