IJE Advance Access originally published online on June 27, 2007
International Journal of Epidemiology 2007 36(4):931-932; doi:10.1093/ije/dym127
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Letters to the Editor |
Metaphorical measurements and theories
The Health Policy Center. 6704 Barr Road, Bethesda, MD 20817, USA.
E-mail: gorigb{at}msn.com
A recent letter of mine asked what consideration should be given to individual recalls of lifetime exposures and experiences: a question that is central to the relevance of most modern epidemiology.1 In their reply, Drs. Fox, Lash and Greenland mistook the question and addressed the different issue of exposure misclassification and other binary biases.2 Rather, the raised problem focuses on the quantitative measurement of the primary variables tested, and on the resulting continuous data such as: how much was each individual exposed over a lifetime to specific dietary items, to occupational and environmental toxicants, to medicines, to second-hand tobacco smoke, and the like. The bulk of these data is obtained from personal recalls of individual study participants or from recall proxies from next of kin of deceased subjects.
Epidemiologists traditionally interpret such recalls as ‘measurements’. Yet, by any factual standards—scientific or otherwise—this qualification is not sustainable, because the sine qua non of statistical elaborations is that discrete characteristics of individuals must have been physically measured using the same meter, that the measurement error is known from prior testable experience, or that it has been determined in the study at hand by multiple measurements of each characteristic, on a sufficient number of subjects.
The interpretation of recalls as ‘measurements’ is grounded on the illusory assumption that individual recalls represent the integral sums of actual measures of myriad momentary and variable exposures over lifetimes, that such measures have been taken with the same metric by all respondents including the next of kin, and that the error of individual recalls is random and can lead to reliable central values for sufficiently large samples. In reality, memories are notoriously metaphorically constructive,3,4 and attempts at their validation by repeated recalls have been disappointing.5–8 Even if repeated recalls were consistent, they still would not represent actual measurements, having no demonstrable connection to physical assessments, not to mention the problem of dimensional heterogeneity raised in the initial letter.1
For added reserve, exposure alone is not a tenable proxy of the ultimate information required, namely the actual doses internalized by each individual.
Recall uncertainties are also overlaid by sampling errors, confounders, biases, misclassifications and more. Such difficulties can only be subjectively tamed by current methods, which—as Drs Fox, Lash and Greenland seem to agree—are generally grounded on the investigator's opinion of the validity of the methods used, on assumptions about the sources of uncertainty and on selections of sensitivity and specificity distributions, all of which remain subjective even when driven by explicit judgement.
In earlier decades, recall-derived epidemiologic reports were correctly interpreted as hypothesis generators. They led to multiple, massive, and long lasting population intervention and clinical trials, designed to test the most promising epidemiologic hypotheses. It is a barely kept secret that, with a notable exception for cigarette smoking, decades of exceedingly costly experimentation have spectacularly failed to confirm the original hypotheses.1
Indeed, whether the uncertainties of recall data could be remedied will determine if much of current epidemiology remains on the dreaded sidelines of science. A suggestion of remedies is tempting, but would not be sanguine until the problem is generally acknowledged and debated. As a step in the right direction, it seems desirable that people should decide to be first epidemiologists and then statisticians, for the construction of elaborate statistical monuments applied to phantom data could only lead to what the editor of this journal has called: ‘the illusory comfort of perhaps metaphorical meta-theories that appear to explain everything while accounting for nothing’.9
It is not a matter of limited academic consequences, because recall-derived epidemiologic reports are blindly taken by policymakers and activists as representing solid scientific prescriptions—prescriptions that are then forced to inconvenience billions of people without valid assurances of benefit, and very possibly with harmful effects. By any honourable account, as epidemiologists we are duty bound to offer some explanation.
Conflict of interest: None declared.
References
1 Gori GB. Dimensional errors of metaphorical measurements. Can they be resolved? Int J Epidemiol (2006) 35:1590–92.
2 Fox MP, Lash TL, Greenland S. Response to ‘dimensional errors of metaphorical measurements: can they be resolved?’ Int J Epidemiol (2006) 35:1592–93.
3 Draaisma D. Metaphors of Memory. A History of Ideas About the Mind (2000) Cambridge, UK: Cambridge University Press.
4 Schacter DL, Addis DR. Constructive memory: the ghosts of past and future. Nature (2007) 445:27.[CrossRef][Medline]
5 Flagg EW, Coates RJ, Calle EE, Potischman N, Thun MJ. Validation of the American Cancer Society Cancer Prevention Study II Nutrition Survey Cohort Food Frequency Questionnaire. Epidemiology (2000) 11:462–68.[CrossRef][Web of Science][Medline]
6 Kroke A, Klipstein-Grobusch K, Voss S, et al. Validation of a self-administered food-frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) Study: comparison of energy, protein, and macronutrient intakes estimated with the doubly labeled water, urinary nitrogen, and repeated 24-h dietary recall methods. Am J Clin Nutr (1999) 70:439–47.
7 Pron GE, Burch JD, Howe GR, Miller AB. The reliability of passive smoking histories reported in a case-control study of lung cancer. Am J Epidemiol (1988) 127:267–73.
8 Coultas DB, Peake GT, Samet JM. Questionnaire assessment of lifetime and recent exposure to environmental tobacco smoke. Am J Epidemiol (1989) 130:338–47.
9 Davey Smith G. Learning to live with complexity: ethnicity, socioeconomic position, and health in Britain and the United States. Am J Publ Health (2000) 90:1694–98.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||