The conundrum of height and mortality.

Samaras and Elrick reiterate a claim they have made else-where—thatanimal and human data show that greater body size is related to shorter lifeexpectancy. The confidence of their claim could lead readers to think that itis based on clear and consistent evidence of a positive association betweenheight and mortality rates. Even the evidence they cite, however, does notsupport this association. For example, they describe a study by Liao andcolleagues as finding no relation between height and heart disease and a studyby Okasha et al as finding no association between height and all-causemortality. It is unclear, however, why such studies, which find no association betweenheight and an outcome, support their claim that “larger body sizeindependently reduces longevity.” Also puzzling is why Samarasand Elrick cite these particular studies out of the body of literature onheight and health. When considered in total, this literature indicates that aconsistent inverse association exists between height and all-causemortality in developed countries. This inverse association has consistentlybeen found in prospective studies that appropriately analyze data fromrepresentative population samples. Samaras and Elrick cite one study from our research group (Okasha et al.)as showing no association between height and all-cause mortality. The resultsof the study, in fact, showed an inverse association, although this was notstatistically significant at conventional levels. They fail to cite studieswith greater statistical power in which we showed robust inverse associationsbetween height and all-causemortality.1,2Rather than carry out a systematic review of population-based prospectiveepidemiologic studies, they refer to unrepresentative (and sometimes basicallyuninterpretable) data derived from their own previous reviews (Samaras andElrick and Miller). What is a sensible conclusion to draw from the literature on height andmortality? In developed countries, taller people have lower all-causemortality rates and live longer. The association between height and mortalityfrom specific causes, however, is heterogeneous, with taller people havinghigher mortality rates from cancers that are unrelated tosmoking3 and fromaorticaneurysm.4 The association between height and some cancers may reflect higher levelsof calorie intake during growth in infancy and childhood, which has been shownto increase later-life cancer risk in a variety of animal models. Insulin-likegrowth factors may mediate thisassociation.5 Thepositive association between height and risk of aortic aneurysm may simplyreflect the mechanical correlation of height and aortic length. Alternatively,it could indicate a Marfan syndrome-like tendency, being more common in tallerpeople. But these positive associations are more than counter-balanced by theinverse associations between height and mortality resulting from coronaryheart disease, stroke, and respiratorydisease,1,2at least in developed countries. Such inverse associations are notattributable to reverse causation, ie, illness in later life leading toaccelerated shrinkage and also increased mortality risk. Height measured inearly adulthood, before any shrinkage occurs, is inversely related tocardiorespiratory mortalityrisk.4 Being tallerthan average is an indicator of favorable childhood social circumstances, andearly-life deprivation is related to a higher risk of mortality from coronaryheart disease, stroke, and respiratory disease. Greater stature is alsorelated to better lung function (at least in part for purely mechanicalreasons), which may be protective against some causes of death. There is much to learn about the mechanisms linking height andcause-specific mortality. Gaining greater understanding requires reading andinterpreting the literature as it is, rather than selectively citing the“facts” that support the hypothesis we have already decided istrue. The review by Samaras and Elrick is misleading. A better contributionwould be to conduct, using well-established principles, a systematic review ofprospective studies in which height was measured in definable populations andmortality data were collected prospectively.