Alternatives for logistic regression in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio

BMC Medical Research Methodology

Vânia N Hirakata¹  Aluísio JD Barros¹ 

[1]Programa de Pós-graduação em Epidemiologia, Universidade Federal de Pelotas, Brazil

关键词: statistical models;    robust variance;    prevalence ratio;    Poisson regression;    odds ratio;    logistic regression;    cross-sectional studies;    Cox regression;   

Others  :  1143288
DOI  :  10.1186/1471-2288-3-21

 received in 2003-04-25, accepted in 2003-10-20,  发布年份 2003

【 摘 要 】

Background

Cross-sectional studies with binary outcomes analyzed by logistic regression are frequent in the epidemiological literature. However, the odds ratio can importantly overestimate the prevalence ratio, the measure of choice in these studies. Also, controlling for confounding is not equivalent for the two measures. In this paper we explore alternatives for modeling data of such studies with techniques that directly estimate the prevalence ratio.

Methods

We compared Cox regression with constant time at risk, Poisson regression and log-binomial regression against the standard Mantel-Haenszel estimators. Models with robust variance estimators in Cox and Poisson regressions and variance corrected by the scale parameter in Poisson regression were also evaluated.

Results

Three outcomes, from a cross-sectional study carried out in Pelotas, Brazil, with different levels of prevalence were explored: weight-for-age deficit (4%), asthma (31%) and mother in a paid job (52%). Unadjusted Cox/Poisson regression and Poisson regression with scale parameter adjusted by deviance performed worst in terms of interval estimates. Poisson regression with scale parameter adjusted by χ² showed variable performance depending on the outcome prevalence. Cox/Poisson regression with robust variance, and log-binomial regression performed equally well when the model was correctly specified.

Conclusions

Cox or Poisson regression with robust variance and log-binomial regression provide correct estimates and are a better alternative for the analysis of cross-sectional studies with binary outcomes than logistic regression, since the prevalence ratio is more interpretable and easier to communicate to non-specialists than the odds ratio. However, precautions are needed to avoid estimation problems in specific situations.

【 授权许可】

   
2003 Barros and Hirakata; licensee BioMed Central Ltd. This is an Open Access article: verbatim copying and redistribution of this article are permitted in all media for any purpose, provided this notice is preserved along with the article's original URL.

【 图 表 】

Figure 1.

【 参考文献 】

• [1]Greenland S: Interpretation and choice of effect measures in epidemiologic analyses. American Journal of Epidemiology 1987, 125:761-768.
• [2]Savitz DA: Measurements, estimates, and inferences in reporting epidemiologic study results [editorial]. American Journal of Epidemiology 1992, 135:223-224.
• [3]Nurminen M: To use or not to use the odds ratio in epidemiologic analyses. European Journal of Epidemiology 1995, 11:365-371.
• [4]Thompson ML, Myers JE, Kriebel D: Prevalence odds ratio or prevalence ratio in the analysis of cross sectional data: what is to be done? Occupational and Environmental Medicine 1998, 55:272-277.
• [5]Miettinen OS, Cook EF: Confounding: essence and detection. Am J Epidemiol 1981, 114:593-603.
• [6]Axelson O, Fredriksson M, Ekberg K: Use of the prevalence ratio v the prevalence odds ratio as a measure of risk in cross sectional studies [letter; comment]. Occup Environ Med 1994, 51:574.
• [7]Zocchetti C, Consonni D, Bertazzi PA: Estimation of prevalence rate ratios from cross-sectional data [letter; comment]. International Journal of Epidemiology 1995, 24:1064-1067.
• [8]Osborn J, Cattaruzza MS: Odds ratio and relative risk for cross-sectional data [letter; comment]. International Journal of Epidemiology 1995, 24:464-465.
• [9]Hirakata Vânia Naomi: Alternativas de análise para um desfecho binário em estudos transversais e longitudinais [MSc dissertation]. In Depto. Medicina Social. Pelotas, Brasil, Universidade Federal de Pelotas;; 1999.
• [10]Lee J, Chia KS: Estimation of prevalence rate ratios for cross sectional data: an example in occupational epidemiology [letter]. British Journal of Industrial Medicine 1993, 50:861-862.
• [11]Victora Cesar Gomes, Vaughan JP, Kirkwood Betty R., Martines JC, Barcelos LB: Risk factors for malnutrition in Brazilian children. The role of social and environmental variables. Bulletin of the World Health Organization 1986, 64:299-309.
• [12]Wacholder S: Binomial regression in GLIM: estimating risk ratios and risk differences [see comments]. Am J Epidemiol 1986, 123:174-184.
• [13]Traissac P, Martin-Prevel Y, Delpeuch F, Maire B: Régression logistique vs autres modèles linéaires généralisés pour l'estimation de rapports de prévalences. Rev Epidemiol Sante Publique 1999, 47:593-604.
• [14]Martuzzi M, Elliott P: Estimating the incidence rate ratio in cross-sectional studies using a simple alternative to logistic regression. Annals of Epidemiology 1998, 8:52-55.
• [15]Skov T, Deddens J, Petersen MR, Endahl L: Prevalence proportion ratios: estimation and hypothesis testing. International Journal of Epidemiology 1998, 27:91-95.
• [16]Cox DR: Regression models and life-tables [with discussion]. J R Stat Soc B 1972, 34:187-220.
• [17]Breslow NE: Covariance analysis of censored survival data. Biometrics 1974, 30:89-99.
• [18]Lee J: Odds ratio or relative risk for cross-sectional data? [letter]. International Journal of Epidemiology 1994, 23:201-203.
• [19]Lin DY, Wei LJ: The robust inference for the Cox proportional hazards model. Journal of the American Statistical Association 1989, 84:1074-1078.
• [20]Huber PJ: The behavior of maximum likelihood estimates under non-standard conditions. In Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. Vol. 1. Berkeley, CA, University of California Press; 1967:1, 221-233.
• [21]Clayton David, Hills Michael: Statistical Models in Epidemiology. New York, Oxford University Press Inc.;; 1996:367.
• [22]McCullagh P, Nelder JA: Generalized linear models. 2nd edition. New York, Chapman and Hall;; 1989.
• [23]Breslow NE: Generalized linear models: checking assumptions and strengthening conclusions. Statistica Applicata 1996, 8:23-41.
• [24]Lee J: Estimation of prevalence rate ratios from cross-sectional data: a reply. International Journal of Epidemiology 1995, 24:1066-1067.
• [25]StataCorp.: Stata Statistical Software: Release 7.0. College Station, TX, Stata Corporation; 2001.
• [26]Victora CG, Barros FC, Halpern R, Menezes AM, Horta BL, Tomasi E, Weiderpass E, Cesar JA, Olinto MT, Guimaraes PR, Garcia MM, Vaughan JP: Estudo longitudinal da populacao materno-infantil da regiao urbana do Sul do Brasil, 1993: aspectos metodologicos e resultados preliminares. Revista de Saúde Pública 1996, 30:34-45.
• [27]Deddens J, Petersen MR, Lei X: Estimation of prevalence ratios when PROC GENMOD does not converge. In Proceedings of SAS Users Group International 28 (SUGI28). Seattle, Washington; 2003:Paper 270.
• [28]Fonseca SS, Victora CG, Halpern R, Barros AJ, Lima RC, Monteiro LA, Barros FC: Fatores de risco para injúrias acidentais em pré-escolares. Jornal de Pediatria 2002, 78:97-104.
• [29]Mendoza-Sassi R, Beria JU, Barros AJ: Outpatient health service utilization and associated factors: a population-based study. Revista de Saúde Pública 2003, 37:372-378.
• [30]Hallal PC, Victora CG, Wells JC, Lima RC: Phisical inactivity: prevalence and associated variables in Brazilian adults. Medicine and Science in Sports and Exercise 2003., in press