Memories and Lessons from The Fels Longitudinal Growth Study

Yellow Springs, Ohio, USA 45387-1695

The Fels Longitudinal Study (Fels Study) is not typical of longitudinal growth studies because it has continued longer than others, it is a study of families that extends into late middle age, and it addresses serious public health concerns in addition to growth. This study began in 1929 and was conducted within the Fels Research Institute, which became part of the nearby School of Medicine at Wright State University. Annually about 15 participants are enrolled into the study before birth from local families that differ in socio-economic status. There are many familial relationships among the participants who are examined 5 times to 1 year, each 6 months from 1 to 18 years, each 2 years from 18 to 24 years and then at 5-year intervals. The variables presently recorded relate to size, maturity, body composition, risk factors for disease, genetic markers, grip strength, physical activity and data from radiographs of the hand. Data relating to many other topic areas were collected in the past.

It was a good decision to locate the Fels Study in Yellow Springs (population 4,500). This village population is more stable in place of residence than is usual in the U.S. Nevertheless, there were some disadvantages. Each Department in the Institute often lacked a critical mass of scientists and it was difficult to obtain effective help from specialists such as biostatisticians. But the study was, and is, productive. More than 800 articles in peer-reviewed journals and book chapters have been published excluding those concerning behavioral and biochemical topics.

Since the Study began, there have been major changes in the age range of the examinations and in the selection of variables to be measured. A major change occurred in 1976 when total body composition, risk factors for disease, hearing ability and noise exposure were added and the behavioral studies were discontinued. This draws attention the need for a conservative approach to the management of a long-term serial study. New variables should be added only if they are biologically related to those already being collected and if their collection will not be too burdensome for the participants. On the other hand, variables being collected should be discontinued when there is doubt about their usefulness. Replicate data are collected and analyzed routinely and close attention is given to the calibration of equipment. The data are stored electronically, but database management is complex because there are many variables and ages at examination and a need to document changing familial relationships.

A longitudinal study presents many opportunities, most of which depend upon effective analyses of serial data. This may involve the development of parametric models for skeletal variables for which the patterns of growth are similar among individuals despite differences in intercepts, rates of growth, and the timing of critical events. Variables derived from such functions can be used in subsequent analyses. Other analyses can be made of the prediction of adult stature and tracking across age can be investigated.

Familial and genetic analyses are possible in the Fels Study because there are more than 1000 participants with data from infancy and cross-sectional data on more than 2000 of their primarily adult relatives. These are grouped in about 200 kindreds consisting of both nuclear and extended families. Much early work in this area was descriptive; later familial correlations were reported. Such studies can provisionally identify traits that may be heritable and may be inherited in a Mendelian fashion. It was shown that there are high heritabilities for the parameters of functions fitted to recumbent length in infancy and to stature in older children. There are incomplete pleiotropic effects of genes for stature indicating that different growth curve parameters may have unique genetic bases. Other analyses showed pleiotropic effects of genes on Body Mass Index and blood pressures that become larger after pubescence.

During any long-term serial study there are numerous difficulties and challenges. These will be considered in detail in relation to recruitment and retention, the influence of secular changes and the balance between continuity and the adoption of improved methods of measurement. Another major aspect relates to funding. It is possible to obtain funding to begin long-term studies. The related problems vary from one country to another. The study need not be as long as that at Fels and the scope need not be as broad. In many cases, a short-term serial study should be initiated including a genetic component. If this study is successful, funding may become available for its continuation. At least in the U.S., funding success depends on the submission of a design with the potential to answer public health questions that cannot be answered more quickly and cheaply from simpler studies. Funding agencies must be convinced that the design is appropriate and they are likely to give particular attention to the leadership of the study. The major key to continuation of funding is the frequent publication of serial analyses in peer-reviewed journals.

The future of long-term studies will be discussed with reference to the inclusion of new methods of data collection and collaboration among studies. There is a need to use established and improved statistical methods, including multivariate serial analyses. While complex statistical procedures are commonly needed, the results should be presented in ways that assist understanding and facilitate their application.

Other analyses, at least in the Fels Study, will locate and identify quantitative trait loci that influence the changes observed in serial data. For this purpose, high-resolution genomic marker maps will be constructed for each participant and linkage analysis methods will be employed to scan for marker variants linked to values at specific ages.