22 Methods of studying human growth

Dr. Heemanshu Aurora

Learning objectives

• This module will focus on different methods employed in studying growth pattern among population groups.
• It will elaborate on the advantages and disadvantages of studying growth of children in all age groups following cross sectional, longitudinal, semi longitudinal, linked longitudinal and Mixed longitudinal design.

INTRODUCTION

A child’s growth rate reflects, his state of health and nutrition better than any other single index. The average value of children’s heights and weights reflect accurately the state of a nation‘s public health (Eveleth and Tanner).

The relevance of human auxology as predictor of health status of individual as well as population as a whole has evolved considerably over the years. It involves concepts, which are closely related to the processes governing today’s rapidly changing bio-social environment. It is well known that child’s net growth at a given point of time is a resultant of interaction between genetic and environmental factors. The childhood environment and growth in particular, often offered as possible predictors of health outcomes and longevity (WHO, 1990; Stini, 1978; Gertler et al., 1951) have scarcely been studied perhaps for want of either necessary infrastructure or use of proper and precise methodology.

For proper understanding of concepts like maximal growth may not be equal to optimal growth (Stini, 1978; Walker, 1983). Also to explore if adult stature is a proxy measure for growth during childhood, the relationship of disease and other growth affecting factors should be more readily demonstrable by directly relating growth parameters to later health outcomes. Hence to understand aforesaid issues it becomes imperative on the part of health professionals, auxologists, anthropologists and nutritionists to use precise methods of studying human growth to arrive at meaningful conclusions, rather than merely quantifying the extent of change that occurs in different body parts of a growing child.

During growth examination surveys ,children investigators are more concernedwith mean values and accompanying different measures of variability statistics pertaining to any of the body parameters measured amongst groups of children rather than with the nature of growth pattern exhibited by them. Therefore to obtain adequate sets of age and sex specific data to be used for comparative evaluation researchers most commonly adopt cross- sectional or longitudinal approach to carry out research in the field of human growth. In between a mixed longitudinal or semi- longitudinal or linked –longitudnal method that represents short term limited application of the longitudinal technique or an extension of the cross sectional approach is often employed (Garn and Shamir,1958).

The first longitudinal studies

A son and some students were the first to be studied longitudinally.The Count du Montbeillard of France measured the stature of his son every sixth months from the boy’s birth in 1759 to his  eighteenth birthday. These data are usually considered to constitute the first longitudinal study of human growth. This was thus the first example of a distance curve of growth (Buffon 1777).

Another eighteenth century longitudinal study of growth is that of the students of the Carlschule, conducted between the years 1772 and 1794. The growth data showed that the sons of the nobility were, on average, taller than the sons of the bourgeoisie during the growing years, but that both groups achieved an approximately equal height at 21 yrs of age. Thus, the sons of the nobility experienced an advancement of the rate of growth (Komlos 1992).

In 1835, Lambert Adolphe Quetelet (1796–1874) published the first statistically complete study of the growth in height and weight of children. Quetelet was the first researcher to make use of the concept of the ”normal curve” (today commonly called the normal distribution or ”bell-shaped” curve) to describe the distribution of his growth measurements, and he also emphasized the importance of measuring samples of children, rather than individuals, to assess normal variation of growth.

Statistical approach of Quetelet was followed in Europe by Luigi Pagliani (1847–1932). He began his studies on the size and fitness of Italian military personnel. He later applied his methods to children and demonstrated that the growth status of orphaned and abandoned boys, ages 10 to 19 y, improved after they were given care at a state-run agricultural colony. Pagliani also noted that children from the higher social classes were taller, and heavier, than poverty-stricken children. Finally, he was taking longitudinal measurements of the same children. From these Pagliani noted that menarche almost always followed the peak of the rapid increase in growth that takes place during puberty (Pagliani 1876).

The early, big studies were cross-sectional

Villerme’ (Villerme’ 1829), Chadwick (Chadwick 1842) and Bowditch (Bowditch 1879), and from Scandinavia Wretlind (Wretlind1878), Vahl (Vahl 1874–83), Key (Key 1885), Malling-Hansen (Malling-Hansen 1883), and further Schmidt-Monnard (Schmidt-Monnard 1895), Boas (Boas 1892), Camerer (Camerer 1893), and Dovertie (Dovertie 1895) all made surveys on the growth of infants, children or adolescents and were the pioneers in this field. Bowditch was the first to perform mass investigations of the development of growth in different ages, and in Boston 13 691 boys and 10 904 girls were measured (height and weight) in different schools.

Boas’ scientific discoveries also include his research into the methodology of growth studies (Boas 1892, Boas 1930). Among other things Boas provided the concept of tempo of growth to understand the difference between early and late maturing individuals.

CROSS SECTIONAL METHOD

A cross sectional method of studying human growth involves measuring children only once during the entire tenure of the study. For example in this method a group of children being measured at 6.0 years of age would be altogether different from those who might have been measured a year ago. Or whom one might be proposing to measure later at the age of 7.0 years. Hence in this method there is no element of periodic assessment. Cross sectional surveys provide information about distance growth attainments of different body parts of the children and reveal enough about the basic nature of the distance growth curve of any anthropometric parameter.

In making cross sectional standards the numbers of children measured in each age group should be approximately proportional to the rate at which growth is taking place. Thus about three times as many children should be measured in the first year after birth as at age 8-9 years, for example put another way, the first year should be broken up into three periods (0-0.32year,0.33-0.66 years,0.67-0.99 years) in each of which same numbers of children should be measured as in the middle childhood year. The second year should be broken into two periods (1.00 – 1.49 years, 1.50-1.99 year). Major standardising studies sample about 1000 boys and 1000 girls in each age group but half this number produces useful centiles. Samples of less than about 300 in each age group are unsuitable as cross- sectional standardising series. (Eveleth and Tanner, 1976)

ADVANTAGES

Cross sectional method of studying growth is cost effective and less time consuming.

It involves much larger number of subjects and provides good deal of information about distance attainments (gross- size) of children (e.g.,10kg of body weight and 76.0cm of supine length at 1 year of age).

This method is a rapid way of assessing magnitude of growth or nutritional deficits and other health problems prevailing amongst children representing different communities at any given point of time. In this method large number of subjects could be measured in a brief duration thus, data generated through cross sectional surveys are often considered as better representative of a population and may be used to establish cross sectional growth standards.

DISADVANTAGES

Owing to their basic inherent nature cross-sectional studies have one chief drawback: they can never provide information about differences in rate of growth (i.e. growth velocity) of different body parameter, as they remain devoid of periodic follow- ups of children. Actually, it is these differences which provide insight into the dynamics of growth of children known to be controlled by genetic, variety of environmental, nutritional, psychological, socio-cultural and economic factors.

Another limitation of cross-sectional data is,it does not precisely reveal about timing of occurrence of some particular phases of growth of children, e.g., onset of ‘juvenile growth spurt’, adolescence and attainment of ‘peak height velocity’, etc. in clear terms, as cross sectional studies do not involve repeat examinations of children so growth velocity related information obtained from such data misleads to arrive at faulty decisions. Therefore as a caution, computation of ‘growth velocity’ of anybody parameter should never be made from cross-sectionally gathered sets of data.

The cross-sectional surveys being one- age point specific in nature reflects conditions only during one season of one particular year, thus provide limited opportunities to have complete insight into the process which affect growth or nutritional status of people and their life style.

Thus to obtain such standards we need a longitudinal study.

LONGITUDNAL METHOD

In a longitudinal method of studying human growth each child enrolled in the study is periodically measured for one or more body parameters at fixed time intervals throughout the study span. For e.g. all children measured at age 6.0 years will remain as the same as those who were earlier examined at age 5.0 years. The constancy of sample size and strict adherence to stipulated periodicity at which children are to be followed up remain the prime requisites of this method. A growth study may be longitudinal for any number of years. It may extent post-natally from birth to even 20 + years. However in actual practice because of variety of reasons it may not be possible to measure same group of children over a prolonged duration of time.

To obtain the simplest type of velocity standards individuals have only to be measured twice. This is what was done in the large scale London standardising study of 1996- 7 and the Cuban National Study of 1972-4.More intensive longitudinal studies follow the children over longer periods. The investigator then has to fit a growth curve to the serial measurements of each individual. For many purposes graphical fits (Tanner et al.1966) are sufficient, but mathematical curves may also be employed (e.g Marubini, Resele, Tanner and Whitehouse,1972).

The major drawback of a comprehensive longitudinal study is the time it takes to complete and the relatively small number of subjects that can be followed. To overcome these problems Eveleth and Tanner (1976) suggested a good design of ‘Linked Longitudinal studies’.

LINKED LONGITUDNAL METHOD

This method studies covering the ages of 0 to 6, 5 to 11, 10 to 15, 14 to 20 years through this design within a period of about 6 years whole age range of growing phase of human life could be spanned by considerably maintaining longitudinal element of the study. However, to obtain smooth joins of the data collected during short intervals an efficient sampling of the population has been advocated by these authors. So a careful initial sampling design requires to be followed and subsequently adhered to so as to obtain suitable sets of distance and velocity growth related data on children growing under a variety of urban , rural and other different environmental and socio- economic circumstances.

For an intensive investigation of the relation between continuously unfolding events in individuals and very often for clinical investigation of growth disorders, long-term longitudinal studies, even from birth to maturity are necessary. For some outcomes types of problem follow-up studies suffice, wherein a cohort is studied at relatively few ages (Eveleth and Tanner 1976). The British National Child Development Survey is an example, wherein a national sample of all children born in one week of March 1958 was studied again at ages 7, 11, and 15(Davie, Butler and Goldstein, 1972; Goldstein, 1971). Depending upon the objective, which one wants to accomplish results from short term longitudinal data can also be obtained, where only some phases of life (i.e., infancy, pre-school years, pre-adolescence, adolescence) could be studied even at larger age intervals (6 months or yearly), but with reasonable accuracy.

ADVANTAGES

Longitudinal studies besides, providing information about the distance (gross-size) growth attainments also provide growth velocity related data which offer true insight about the growth rates velocity of different body parameters with which these change periodically over a period of time. Such changes in the different body parts of a growing child may be incremental or decremental in nature. As the growth velocity denotes innate capacity of a child to grow and develop so these studies help to understand the influence of various environmental, nutritional, seasonal variations, disease and climate related factors on growth dynamics of children.

Most illnesses in childhood are short – termed. Longitudinal studies provide opportunity to have information about the duration of any disease or nutritional deficiency with which any child might have remained afflicted with for a longer period. Similarly, effect of any intervention (medical/nutritional etc.) may also be assessed precisely through these studies as children included in such surveys are often monitored periodically.

DISADVANTAGES

Longitudinal growth studies are very expensive and difficult to organize.

Lot of time and effort are consumed to carry out, analyze and interpret voluminous data, which are often generated over the years. These studies require lot of motivation, commitment as well as patience on the part of both subjects as well as researchers to be present on the stipulated date and time of examination during the entire duration of the study, which in fact for obvious reasons is not possible for a very long duration.

Longitudinal studies involve examination of limited number of children. It is not always possible to maintain consistency of sample size throughout the entire span of study.These studies exhibit substantial sample attrition rate. Many subjects have to leave the study as they move to new places mainly because of social as well as occupational reasons.

HAWTHRONE EFFECT: This is unique but not commonly known phenomenon in the context of human growth .The principal drawbacks of longitudinal studies are the time these take to complete and the small number of subjects that can usually be followed up. One of the ways to maintain adequate sample size in such studies is to generate a high level of involvement amongst children participating in the study. The phenomenon of greater involvement of parents in a longitudinal series yields to Hawthorne Effect (Blalock, 1970) which results from subjects knowing that they are a part of study. According to Johnston (1980), continuous and persistent health care advice to the parents (whose children are a part of longitudinal study) may become aware of relationship between growth and need for proper environment. As a result they may change their children’s diet included in the longitudinal study.

The association of Hawthorne effect with longitudinal studies was done by Bhalla and Kumar (1986). They compared weight and crown- heel length of 154(86 males; 68 females) longitudinally studied infants at monthly intervals with those measured amongst721 (365 males; 356 females) taken cross sectionally. Infants representing longitudinal study demonstrated improved general growth status in terms of both weight and crown- heel length measurements in contrast to those belonging to cross sectional series beyond 4to 5 months of age by which parents often start responding to various components of health care advise (supplementary feeding, timely treatment of illnessesetc). which is often given to them when they come for routine check-ups. This improved growth status has been attributed to Hawthorne Effect as every child included in the longitudinal series was examined for 13 times in a period of one year. Whereas, such situation was absent in cross sectional series, which always lack repeat examinations.

From a methodological point of view, Hawthorne effect appears to be an inbuilt and inevitable feature of longitudinal design of studying Human growth. At one end it leads to the improvement of general growth status of children included in the longitudinal study solely because of the basic nature of the design. Whereas on the other hand it looks like harsh indictment of the longitudinal method, as infants with improved growth status stop representing the population from which they were originally drawn as sample. Above comments are not intended to convey that longitudinal design is not basic to the study of growth ,rather ,it is to emphasise that a longitudinal study should be initiated only after a careful consideration of all aspects of design, of which HAWTHORNE effect could be an important one(Bhalla and Kumar,1986).

MIXED LONGITUDNAL METHOD

A serial study in which a group of children is followed such that ‘some children leave the study and others, if desired join it’ (Tanner, 1977), as new entrants at some age points is termed as mixed-longitudinal study.

It is not always possible to maintain constancy of sample size particularly, when a longitudinal study is to be carried out for a number of years. Usually, in populations which show more of mobility, children often accompany their parents and leave the villages or towns or cities etc. Many a times some children because of unwillingness stop participating and others particularly, during adolescence feel reluctant to get examined for sexual maturity ratings etc. and often do not report for examination. In addition to these, many a times, some local, seasonal, personal and socio-economic and health related reasons come into play and children fail to report for investigations. The reasons may be of permanent or temporary nature but this effect the sample at different age points. On one end the variable sample attrition rate experienced by subjects enrolled in a mixed longitudinal study due to variety of aforementioned reasons results in accumulation of data with missing values ,but on the other hand this design provides the opportunity to make up for the sample loss, if desired, by enrolling new subjects at any requisite age – points.

Mixed longitudinal studies are relatively cheaper to conduct, and are also less time and effort consuming as compared to pure longitudinal series. These studies like pure longitudinal studies provide us with both distance and velocity growth related data with reasonable authenticity. However, estimation of growth velocity of different body parameters from mixed longitudinal data involving missing values is a tedious task and special statistical methods are required to get maximum information out of such data. To extract out maximum distance and growth velocity related information from mixed –longitudinal series that is one in which children enter and leave the study at different ages, giving variable degree of longitudinally, the use of statistical methods given by Tanner (1951) and Goldstein(1987) are found to be most efficacious.

In the past, it has been noticed that in most of the instances researchers have treated their longitudinal data.

Semi longitudinal study

This is basically a longitudinal study where all the subjects of the study cannot be followed up for the duration of the study due to logistic difficulties. For example, we are recording growth in 100 children from 0to 5 years. At the first observation, 100 children are recorded. At the second observation, there is a dropout of 10 cases, and we record 90 cases. At the subsequent visits, the numbers of dropouts increase, so that at the end of the study at 5years the numbers of children who have been included in all observations are only 30. So from this study we have a pure longitudinal study in only 30. But the relevant data also from the rest 70. In these 70 children we can, by using appropriate mathematical devices convert these data into a velocity curve. In another type of mixed longitudinal data, the cohort changes with progress of time. If at the outset we have 40 cases to start with, the cohort might build up to 50 cases at the second occasion and then with further dropouts reduce to 30 at a subsequent occasion. So data can also be incorporated into a velocity curve (A. Parthasarathy, 2007)

This type of study is a compromise between the cross sectional and pure longitudinal studies and we can have some information like longitudinal data besides what is obtained in a cross sectional study. These studies are less expensive and less time consuming compared to pure longitudinal studies. They help us in getting efficient estimates, with minimum variance of average at points, on increments, velocities and acceleration of growth for the different anthropometric measurements for comparison and to detect any abnormality of children during their growth period.

SUMMARY

By the very nature of their field of study, health professionals, auxologists, anthropologists are concerned with change that occurs over time. This fact brings to light the need to choose research design to arrive at meaningful conclusions, rather than merely quantifying the extent of change that occurs in different body parts of a growing child. So far five methods have been employed in growth studies these are Longitudinal method, Cross- sectional method, linked longitudinal method, Mixed longitudinal and Semi longitudinal method. Longitudinal studies involve studying the same group of participants over a particular time period. Cross-sectional studies involved studying groups of participants in different age groups at the same point in time. To overcome the disadvantages of longitudinal studies Eveleth and Tanner (1976) suggested a good design of ‘Linked Longitudinal studies’. Covering the ages of 0 to 6, 5 to 11, 10 to 15, 14 to 20 years through this design within a period of about 6 years whole age range of growing phase of human life could be spanned by considerably maintaining longitudinal element of the study. A Mixed longitudinal studies a one in which a group of children is followed such that ‘some children leave the study and others, if desired join it’ (Tanner, 1977), as new entrants at some age points .Semi longitudinal study is a compromise between cross sectional and pure longitudinal studies.