Post on 04-Jan-2016
description
EPID 623-88 Introduction to Analysis and
Interpretation of HIV/STD Data
Confounding
Manya Magnus, Ph.D.Summer 2001
adapted from M. O’Brien and P. Kissinger
Definition of Confounding
• A non-causal association between a given exposure and an outcome is observed as a result of the influence of a third variable (or group of variables) designated as confounding variable(s).
Rules of Confounding
• The confounding variable is:– Causally associated with the outcome– Non-causally or causally associated with
the exposure– Not an intermediate variable in the causal
pathway between exposure and outcome
Types of
• Positive – overestimation of the true strength of association
• Negative – underestimation of the true strength of association
• Qualitative – inverse in the direction of the association
Different strategies to assess confounding
• Examine crude and adjusted estimates of the association
• Stratification and examination of measures of association by strata
Crude Associations
More ideas about confounding
• Partial confounding can occur (not an all or nothing thing)
• Residual confounding (occurs when categories of confounders controlled for are too broad or when confounding variables remain unaccounted for)
Collinearity
Effect Modifiers
Interaction
• Two or more risk factors modify the effect of each other with regard to the occurrence or level of a given outcome
• Also known as effect modification• Synergistic (positive interaction) – potentiates
the effect of the exposure of interest• Antagonistic (negative interaction) –
diminishes or eliminates the effect of the exposure of interest
Confounding versus Interaction
• Sometimes the same variable may be both a confounder and an effect modifier
• Confounding makes it difficult to evaluate whether a statistical association is also causal
• Interaction is part of the web of causation• Do not adjusted for a variable that is both a
confounder and an effect modifer (reporting an average odds may be meaningless)
Risk factors for sinusitis among HIV-infected
persons in Multivariate logistic regression Sinusitis
(n = 521) No
Sinusitis (n = 3645)
Unadjusted O.R. Adjusted O.R. (95% C.I.)
Race White Non-white
42.6 57.4
32.5 67.5
1.54 (1.28 – 1.86)** 1.00
1.67 (1.37 – 2.04)** 1.00
Age <35 35
49.3 50.7
44.9 55.1
1.19 ( .99 – 1.43) 1.00
1.19 ( .99 – 1.44) 1.00
Sex Female Male
29.6 70.4
26.9 73.1
1.13 ( .93 – 1.39) 1.00
1.30 (1.04 – 1.62)* 1.00
CD4 at entry 200 < 200
64.3 35.7
59.1 40.9
1.24 ( 1.03 – 1.51)* 1.00
1.33 (1.07 – 1.65)** 1.00
Opportunistic Infection Present Absent
29.2 70.8
25.9 74.1
1.18 ( .96 – 1.44) 1.00
1.43 (1.34 – 1.80)** 1.00