Linear Growth Faltering Should Be Assessed in Absolute and Relative Terms

Dear Editor:The concept of standard deviations was rst proposed by KarlPearson 120 y ago (1) to account for the fact that an absolutemeasure (e.g., in centimeters or kilograms) may have differentinterpretations depending on how much variability there iswithin a population. Had Pearson still been among us, he wouldnot be surprised by the ndings of Leroy et al. (2)—namely,that conicting results are produced when growth faltering isassessed in absolute (centimeters) or relative (z scores) scales.Leroy et al. reanalyzed data from 51 national surveys in low-and middle-income countries included in our 2010 publication onthe timing of growth faltering (3). Consistent with our ndings,they concluded that there was substantial faltering during the rst2 y of life. Thereafter, mean height-for-age, expressed as z scoresor SDs of the WHO Child Growth Standards (HAZ), remainedstable at a z score of approximately 21.8upto5yofage.Inthesame age range, however, they found that the mean absoluteheight-for-age decit (HAD) increasedby ;2cm(Figure1intheirarticle). Their extensive reanalyses were not quite necessary.Simple arithmetic applying our 2010 results to the WHO ChildGrowth Standards will reveal that the mean decit of a 21.8zscoreat24mocorrespondsto 25.5cm,whereasthesame zscorevalue corresponds to 28.2 cm at 60 mo (averaging boys and girls),adifferenceof2.7cm.In general, their article is well balanced. Their main conclusionsdonotdetractfromthoseinouroriginalpublications(3,4),namelythattherst1000dconstitute aunique windowof opportunityfornutritionalinterventions.However,wemusttakeexceptiontotheirsweepingconclusionthat“HAD, ratherthanHAZ,shouldbeusedtodescribeandcomparechangesinheight.”Takingtheirargumentto its extreme, a 2-cm decit for 24-mo-old boys ( 20.65 z score)would be equivalent to a 2-cm decit in 19-y-old men (2 0.27z score) (5), and such a population would have remained stableovertimeintermsofheightdecit.Onemightjustaswellarguethatthere was denite catch-up in relative (SD) terms, and that being2cmshortofthemedianasanadultispreferabletobeing2cmshortat the age of 2 y.There are many instances in biology and in public health inwhich absolute and relative scales provide apparently contra-dictory results. For example, if stunting prevalence in group Adecreases from 40% to 20% and prevalence in group Bdecreases from 10% to 5%, the absolute gap is reduced from30 to 15 percentage points, but the relative measure (prevalenceratio)remains4-fold.Choosing1measureovertheotherimpliesa value judgment (6), and there is no single correct answer.We argue that expressing growth faltering in absolute termsprovides an alternative, complementary approach to the relativemeasures used in our earlier publications. Each approach hasa different interpretation, and both are worth presenting (7).It is not only a matter of when growth falters. It is alsoa matter of which tissues and organs grow more rapidly duringdifferent age ranges, such as the well-documented rapid braingrowth in the rst 2 y (8). Emphasis on the rst 1000 d is basednot only on the magnitude of faltering but also on its long-termimpact on adult human capital, as well as on the fact thatpromoting rapid weight gain after this critical window has beenstrongly associated with the risk of noncommunicable diseasesin adulthood (9,10). Although rapid linear growth after the ageof 2 y does not appear to lead to increased noncommunicabledisease risk (11), at the present state of knowledge one cannotbe sure that promotion of linear growth will not also lead toexcessive weight gain. Therefore, although from an academicstandpoint both absolute and relative faltering are noteworthy,from a programmatic standpoint focusing on the correction ofabsolute decits after 2 y of age may be detrimental.Cesar G Victora*Postgraduate Program in EpidemiologyFederal University of PelotasPelotas RS, BrazilMercedes de OnisDepartment of NutritionWHOGeneva, SwitzerlandRoger ShrimptonDepartment of Global Community Health andBehavioral SciencesSchool of Public Health and Tropical MedicineTulane UniversityNew Orleans, LA