On Analysis of the Difference of Two Exposure-Adjusted Poisson Rates with Stratification: From Asymptotic to Exact Approaches
- First Online:
- 220 Downloads
Long-term studies are frequently conducted to assess the treatment effect on rare diseases or the safety of a new treatment. To account for differential follow-up often encountered in long-term studies, exposure-adjusted incidence rates are used in evaluating the treatment effect. The difference of rates is sometimes used to quantify the treatment’s public health impact because the reciprocal of this difference can be interpreted as “the number needed to treat (or number needed to vaccinate) in order to cure (or prevent) 1 case of disease.” In this paper we focus on the stratified analysis of the difference of two exposure-adjusted rates in the setting of superiority, noninferiority and super-superiority hypothesis testing. After a brief review of asymptotic methods, we derive an exact method that guarantees control of the type I error. But it is conservative for noninferiority and super-superiority testing because of the search of the maximum tail probability over a multidimensional nuisance parameter space. Then, we present an approximate exact method where the p-value is estimated at the maximum likelihood estimates of the nuisance parameter. This method is identical to exact method for superiority testing and reduces the conservatism for noninferiority and super-superiority testing. In addition, a quasi-exact method is discussed. A real-life vaccine clinical trial example is used to illustrate these methods. Finally, we compare the performance of these methods via empirical studies and make some general practical recommendations.
KeywordsApproximate exact method Constrained maximum likelihood estimate Exact test Poisson rate Risk difference Stratified analysis
Unable to display preview. Download preview PDF.
- 11.Guess HA, Lydick EG, Small RD, Miller LP (1987) Epidemiologic programs for computers and calculators—exact binomial confidence intervals for the relative risk in follow-up studies with sparsely stratified incidence density data. Am J Epidemiol 125:340–347 Google Scholar
- 21.Oxman MN, Levin MJ, Johnson GR, Schmader KE, Straus SE, Gelb LD, Arbeit RD, Simberkoff MS, Gershon AA, Davis LE, Weinberg A, Boardman KD, Williams HM, Zhang JH, Peduzzi PN, Beisel CE, Morrison VA, Guatelli JC, Brooks PA, Kauffman CA, Pachucki CT, Neuzil KM, Betts RF, Wright PF, Griffin MR, Brunell P, Soto NE, Marques AR, Keay SK, Goodman RP, Cotton DJ, Gnann JW, Loutit J, Holodniy M, Keitel WA, Crawford GE, Yeh SSY, Lobo Z, Toney JF, Greenberg RN, Keller PM, Harbecke R, Hayward AR, Irwin MR, Kyriakides TC, Chan CY, Chan ISF, Wang WWB, Annunziato PW, Silber JL (2005) A vaccine to prevent herpes zoster and postherpetic neuralgia in older adults. New Engl J Med 352:2271–2284 CrossRefGoogle Scholar
- 23.Wang WWB, Xu J, Sutradhar S, Chan ISF (2005) On hypothesis testing and confidence intervals for the difference of two independent Poisson rates. Annual Conference of the Eastern North American Region (ENAR) of International Biometric Society, Austin, TX, March 2005 Google Scholar