A sibship's FLoSS is a good predictor of its true (not yet fully known) exceptionality of survival. Thus, it is useful for selecting long-lived families for studies of genetic and nongenetic factors contributing to longevity. In particular, the FLoSS can identify desirable sibships among families being screened for genetic epidemiologic studies of exceptional longevity. Sibships with a high FLoSS have high ages of the total sibship and of living siblings, as well as high numbers of total and living members.
A high FLoSS can be achieved by the presence of a single individual with extremely long survival (if the other siblings are not short lived) or by the presence of many long (but not extremely)-lived siblings. The modes of transmission of longevity (if any) associated with these types of patterns may differ. For example, the former might reflect a recessive trait; the latter, a dominant. The fact that a high FLoSS value captures both of these patterns is one of its strengths, because it does not exclude longevous families who might not be ascertained by other methods (e.g., setting a single extremely exceptional age for inclusion of individuals or siblings). However, the additional potential heterogeneity of genetic factors captured in high-FLoSS families implies a need for attention to potential differences among subgroups in genetic transmission.
We used the FLoSS to score families defined as members of a single sibship. Although it could be used to score multigenerational groups of relatives, gathering complete information on the parents of old siblings or on widely dispersed groups of near relatives could be challenging. Furthermore, young people contribute little information regarding ultimate survival.
In the Long Life Family Study, we set the minimum FLoSS for a family to be eligible at 7. This threshold was chosen by observing that only 0.2% of the FLoSS sibships of the Framington Heart Study meet this threshold, in contrast to over 30% of the Long Life Family Study screening families and over 40% of families enrolled in the New England Centenarian Study. Thus, families with a FloSS as large as 7 are extremely rare but findable. Calculation of FLoSS scores in families from additional population-based samples can provide further guidance about appropriate selection thresholds.
The FLoSS is the sum of
(the living bonus). Its first component is a current estimate of a family's ultimate, “fully observed,” survival exceptionality, SEf
. Although the living bonus enables the FLoSS to select
families that are more desirable for genetic epidemiologic studies of exceptional longevity,
may be a better intrinsic measure of family longevity. As such,
may be particularly useful for examining relations of genotypes and other risk factors to phenotypes of family members and their family's exceptionality of survival, or for finding subsets of populations that are similar with respect to the exceptionality of their families. Although in our comparison the FLoSS and
correlated equally well with the observed SEf
10 years later, this may not apply over longer intervals. Note that, when all family members have died, all 3 measures coincide.
The FLoSS as used in the Long Life Family Study can be viewed as a member of a class of family scores that combine an estimated exceptionality of survival with a bonus for living siblings, as in the following:
is of the form,
are nonnegative constants, C
) is a person-specific reference cohort of those born around the same time who survived to some minimum age, a
, and LBf
is a nonnegative bonus for older living siblings. Each parameter can be tailored to the particular needs of other studies.
For the FLoSS, we chose w
= 1, C
) = all people in the same birth-year, gender, and national cohort who survived to at least age 40, k
= 1, and
An age threshold other than 40 years could be chosen depending on a study's focus (e.g., age 70 to study factors influencing survival only in advanced age). Choosing w
1 weights the estimated survival exceptionality and living bonus equally; w
, a measure of family survival exceptionality alone. Note that the expected value of
for a “pseudofamily” constructed by grouping N
randomly selected individuals from the US population is N
× (1 – k
). Choosing k
neutral to sibship size, because its expected value should be 0 for randomly selected (not particularly long-lived) people. The “neutrality” of the FLoSS was validated by the fact that its average in the geography-based Framingham Heart Study cohort was quite close to 0. Because –ln(0.37) is approximately equal to 1, only ages in the top 37th percentile add to
. More generally, values of k
smaller than 1 favor larger families (because each “typical” person's expected score is positive), while choosing k
larger than 1 favors smaller ones. If greater exceptionality is sought, a larger cutoff for the FLoSS could be used.
We chose w
= 1, giving equal weight to the estimated exceptionality score and the “bonus” for living older family members. This reflected our interest in both survival exceptionality and the availability of old living study subjects. With this choice, the FLoSS still correlates strongly with
as we wanted. Larger w
’s will give more weight to the living bonus and reduce the correlation with
In summary, we have introduced and examined the consequences of a conceptually attractive framework for family longevity studies. These include 1) a measure of an individual's exceptionality of survival, 2) a feasible way to estimate that exceptionality for those still alive, 3) a size-neutral way to combine individual scores into a family score, 4) a plausible bonus measure for the additional value of already exceptional living family members, and 5) a way to balance interest in older living relatives and family survival exceptionality in a single score such as the FLoSS. This framework should be useful in many settings. We also note that formulas for the FLoSS and
could be adapted to measure exceptionality of survival until other events besides death. Thus, we could quantify family risk for the onset of conditions (such as stroke or onset of diabetes or disability) whose incidence rises with age. We are investigating this idea.