Statistical Significance of Geographic Heterogeneity Measures in Spatial Epidemiologic Studies

Min Lian^1,2

¹Department of Medicine, Washington University School of Medicine, St. Louis, Missouri, USA.
²Alvin J. Siteman Cancer Center, Barnes-Jewish Hospital and Washington University School of Medicine, St. Louis, Missouri, USA.

DOI: 10.4236/ojs.2015.51006                

Abstract

Assessing geographic variations in health events is one of the major tasks in spatial epidemiologic studies. Geographic variation in a health event can be estimated using the neighborhood-level variance that is derived from a generalized mixed linear model or a Bayesian spatial hierarchical model. Two novel heterogeneity measures, including median odds ratio and interquartile odds ratio, have been developed to quantify the magnitude of geographic variations and facilitate the data interpretation. However, the statistical significance of geographic heterogeneity measures was inaccurately estimated in previous epidemiologic studies that reported two-sided 95% confidence intervals based on standard error of the variance or 95% credible intervals with a range from 2.5^th to 97.5^th percentiles of the Bayesian posterior distribution. Given the mathematical algorithms of heterogeneity measures, the statistical significance of geographic variation should be evaluated using a one-tailed P value. Therefore, previous studies using two-tailed 95% confidence intervals based on a standard error of the variance may have underestimated the geographic variation in events of their interest and those using 95% Bayesian credible intervals may need to re-evaluate the geographic variation of their study outcomes.

Keywords

Spatial Epidemiology, Heterogeneity, Statistical Significance, 95% Confidence Interval, 95% Credible Interval

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1]	Elliott, P. and Wartenberg, D. (2004) Spatial Epidemiology: Current Approaches and Future Challenges. Environmental Health Perspectives, 112, 998-1006. http://dx.doi.org/10.1289/ehp.6735
[2]	Larsen, K., Petersen, J.H., Budtz-Jorgensen, E. and Endahl, L. (2000) Interpreting Parameters in the Logistic Regression Model with Random Effects. Biometrics, 56, 909-914. http://dx.doi.org/10.1111/j.0006-341X.2000.00909.x
[3]	Larsen, K. and Merlo, J. (2005) Appropriate Assessment of Neighborhood Effects on Individual Health: Integrating Random and Fixed Effects in Multilevel Logistic Regression. American Journal of Epidemiology, 161, 81-88. http://dx.doi.org/10.1093/aje/kwi017
[4]	Merlo, J., Chaix, B., Ohlsson, H., Beckman, A., Johnell, K., Hjerpe, P., Rastam, L. and Larsen, K. (2006) A Brief Conceptual Tutorial of Multilevel Analysis in Social Epidemiology: Using Measures of Clustering in Multilevel Logistic Regression to Investigate Contextual Phenomena. Journal of Epidemiology and Community Health, 60, 290-297. http://dx.doi.org/10.1136/jech.2004.029454
[5]	Chaix, B., Merlo, J., Subramanian, S.V., Lynch, J. and Chauvin, P. (2005) Comparison of a Spatial Perspective with the Multilevel Analytical Approach in Neighborhood Studies: The Case of Mental and Behavioral Disorders Due to Psychoactive Substance Use in Malmo, Sweden, 2001. American Journal of Epidemiology, 162, 171-182. http://dx.doi.org/10.1093/aje/kwi175
[6]	Chaix, B., Rosvall, M. and Merlo, J. (2007) Assessment of the Magnitude of Geographical Variations and Socioeconomic Contextual Effects on Ischaemic Heart Disease Mortality: A Multilevel Survival Analysis of a Large Swedish Cohort. Journal of Epidemiology and Community Health, 61, 349-355. http://dx.doi.org/10.1136/jech.2006.047597