You cannot exclude the explanation you have not considered

Datta, M (1993) You cannot exclude the explanation you have not considered. Lancet, 342 (8867). pp. 345-347. ISSN 0140-6736

[thumbnail of 199306.pdf] PDF
199306.pdf - Published Version
Restricted to Registered users only

Download (32kB) | Request a copy


Analytical epidemiology largely concentrates on the associations between putative risk factors and disease. If an association is found, how can we decide whether the risk is causally related to outcome? In hospital-based studies of children, severe undernutrition is related to serious infections, including severe rather than localised tuberculosis.1 The odds ratios (ORs) are high. Is this relation causal or is there an alternative explanation? The more malnourished are likely to come from poorer families, and financial hardship could lead to undernutrition and late admission, resulting in more severe infection. Variation in socioeconomic conditions could therefore account for some of the association. In a case-control study of risk factors for breast cancer,2 unadjusted univariate analysis suggested that the ORs decreased with parity and rose with age at first full-term pregnancy (FTP) and slightly with age at last FTP. After adjustment for parity, the relation with first FTP became weaker, and that with last FTP stronger. After adjustment of the three factors for each other, the effect of age at first FTP was no longer significant. The true effect of the exposure variables is being masked or altered by one or more factors in the study population that influence development of the disease and are associated with exposure status. This phenomenon is referred to as confounding and the variables as confounders.

Item Type: Article
Subjects: Tuberculosis > Epidemiological Research
Divisions: Epidemiology
Depositing User: Dr. Rathinasabapati R
Date Deposited: 17 Sep 2013 11:04
Last Modified: 14 Mar 2016 06:19

Actions (login required)

View Item View Item