Setting up a multivariable analysis

Mitchell H. Katz

doi:10.1017/CBO9780511974175.007

6 - Setting up a multivariable analysis

Published online by Cambridge University Press: 01 April 2011

Mitchell H. Katz

Show author details

Mitchell H. Katz: Affiliation:
University of California, San Francisco

Book contents

Get access

Summary

What independent variables should I include in my multivariable model?

On the surface this seems like a simple question. You should include the risk factor(s) of interest and any variables that may potentially confound the relationship between the risk factor and the outcome. However deciding which variables may confound your analysis is not always easy. Variables that are extraneous, redundant, have a lot of missing data, or intervene between your risk factor and outcome should be excluded.

Recommendations on what variables to include and exclude in your model are reviewed in Table 6.1 and discussed in the next two sections.

How do I decide what confounders to include in my model?

Ideally researchers should include all those variables that have been theorized or shown in prior research to be confounders. Depending on the outcome you are studying, there may be a large number of variables that have been shown in prior research to be associated with the risk factor and the outcome. For example, studies of cardiovascular outcomes must include a large number of potential outcomes including age, sex, smoking status, hypertension, diabetes, obesity, LDL-cholesterol, HDL-cholesterol, reactive C-protein, aspirin use, and beta-blocker use because all of these variables have already been shown to affect cardiovascular disease.

TIP

Include in your model those variables that have been theorized or shown in prior research to be confounders of the relationship you are studying.

Type: Chapter
Information: Multivariable Analysis
A Practical Guide for Clinicians and Public Health Researchers
, pp. 93 - 117

DOI: https://doi.org/10.1017/CBO9780511974175.007 [Opens in a new window]

Publisher: Cambridge University Press

Print publication year: 2011

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Book purchase

Temporarily unavailable

References

Spencer, F.A., Allegrone, J., Goldberg, R.J., et al. “Association of statin therapy with outcomes of acute coronary syndromes: The Grace study.” Ann. Intern. Med. 140 (2004): 857–66CrossRef Google Scholar PubMed

Golin, C.E., Liu, H., Hays, R.D., et al. “A prospective study of predictors of adherence to combination antiretroviral medication.” J. Gen. Intern. Med. 17 (2002): 756–65CrossRef Google Scholar PubMed

Szklo, M. and Nieto, F.J. Epidemiology: Beyond the Basics. Gaithersburg: Aspen Publishers, 2000, p. 333Google Scholar

Katz, M.H. Study Design and Statistical Analysis: A Practical Guide for Clinicians. Cambridge University Press, 2006CrossRef Google Scholar

Hulley, S.B., Cummings, S.R., Browner, W.S., et al. Designing Clinical Research (2nd edn). Baltimore, MD: Lippincott Williams and Wilkins, 2001, pp. 65–91Google Scholar

Peduzzi, P., Concato, J., Kemper, E., et al. “A simulation study of the number of events per variable in logistic regression analysis.” J. Clin. Epidemiol. 49 (1996): 1373–9CrossRef Google Scholar

Peduzzi, P., Concato, J., Feinstein, A.R., et al. “Importance of events per independent variable in proportional hazards regression analysis II. Accuracy and precision of regression estimates.” J. Clin. Epidemiol. 48 (1995): 1503–10CrossRef Google Scholar

Harrell, F.E., Lee, K.L., Matchar, D.B., et al. “Regression models for prognostic prediction: Advantages, problems, and suggested solutions.” Cancer Treat. Rep. 69 (1985): 1071–7Google Scholar PubMed

Schwarcz, S.K., Katz, M.H., Hirozawa, A., et al. “Prevention of Pneumocystis carinii pneumonia: Who are we missing?” AIDS 11 (1997): 1263–8CrossRef Google Scholar PubMed

Feinstein, A.R.Multivariable Analysis: An Introduction. New Haven: Yale University Press, 1996, p. 226CrossRef Google Scholar

Signorini, D.F. “Sample size for Poisson regression.” Biometrika 78 (1991): 446–50CrossRef Google Scholar

Chambers, C.D., Johnson, K.A., Dick, L.M., et al. “Birth outcomes in pregnant women taking fluoxetine.” New. Engl. J. Med. 335 (1996): 1010–15CrossRef Google Scholar PubMed

Turner, R.J. and Lloyd, D.A. “Stress burden and the lifetime incidence of psychiatric disorder in young adults.” Arch. Gen. Psych. 61 (2004): 481–8CrossRef Google Scholar PubMed

Hull, C.H. and Nie, N.H.SPSS Update 7–9. New York, NY: McGraw-Hill, 1989, p. 257Google Scholar

Glantz, S.A. and Slinker, B.K.Primer of Applied Regression and Analysis of Variance. New York, NY: McGraw-Hill, 1990, pp. 216–36Google Scholar

Kleinbaum, D.G., Kupper, L.L., and Muller, K.E. Applied Regression Analysis and Other Multivariable Methods (2nd edn). Boston, MA: PWS-Kent, 1988, pp. 595–640Google Scholar

Chertow, G.M., Milford, E.L., Mackenzie, H.S., et al. “Antigen-independent determinants of cadaveric kidney transplant failure.” JAMA 276 (1996): 1732–6CrossRef Google Scholar PubMed

Smith, J.L., Rost, K.M., Nutting, P.A., et al. “Impact of ongoing primary care intervention on long term outcomes in uninsured and insured patients with depression.” Med. Care 40 (2002): 1210–22CrossRef Google Scholar PubMed

Heitjan, D.F. “What can be done about missing data? Approaches to imputation.” Am. J. Pub. Health 87 (1997): 548–50CrossRef Google Scholar PubMed

Rubin, D.B.Multiple Imputation for Nonresponse in Surveys. New York, NY: Wiley, 1987CrossRef Google Scholar

Halfon, N., Wood, D.L., Valdez, B., et al. “Medicaid enrollment and health services access by Latino children in inner-city Los Angeles.” JAMA 277 (1997): 636–41CrossRef Google Scholar PubMed

Marascuilo, L.A. and Levin, J.R.Multivariate Statistics in the Social Sciences: A Researcher's Guide. Monterey, CA: Brooks/Cole Publishing Co., 1983, pp. 64–6Google Scholar

Delucchi, K.L. “Methods for the analysis of binary outcome results in the presence of missing data.” J. Consult. Clin. Psych. 62 (1994): 569–75CrossRef Google Scholar PubMed

Little, R.J.A. and Rubin, D.B.Statistical Analysis with Missing Data. New York, NY: Wiley, 1990Google Scholar

Greenland, S. and Finkle, W.D. “A critical look at methods for handling missing covariates in epidemiologic regression analyses.” Am. J. Epidemiol. 142 (1995): 1255–64CrossRef Google Scholar

Kegeles, S.M., Hays, R.B., and Coates, T.J. “The Mpowerment project: A community-level HIV prevention intervention for young gay men.” Am. J. Pub. Health 86 (1996): 1129–36CrossRef Google Scholar PubMed