We also calculated the following crude measures of association comparing ever

# We also calculated the following crude measures of

This preview shows page 3 - 6 out of 6 pages.

We also calculated the following crude measures of associationcomparing ever smokers to never smokers (reference group): CIR = 2.41 over the 22-year study period, 95% confidence interval = (1.69, 1.43) over the 22-year study period CID = 0.0222 over the 22-year study period, 95% confidence interval = (0.014, 0.030)over the 22-year study periodSuppose the data from this study were broken down by age asfollows:<55|>55SmokedNeverTotalSmokedNeverTotalCases9319112721890No-cases2,8601,0853,9451,3211,2222,543Total2,9531,1044,0571,3931,2402,633a. Calculate the cumulative incidence ratio for the associationbetween ever smoking Äb0compared to never smoking and oralor bladder cancer within each stratum of age (≤55, >55 years).
295393/¿¿19¿1104¿¿¿¿¿CIR¿55=¿72¿1393¿¿¿18¿1240¿¿¿¿¿CIR¿55=¿b.Calculate the cumulative incidence difference for the associationbetween ever smoking compared to never smoking and oral orbladder cancer within each stratum of age (≤55, >55 yearÄb0s). 295393/¿¿19¿1104¿¿¿CID¿55=¿72¿1393¿¿¿18¿1240¿¿¿CID¿55=¿c.Do the results from 2a and 2b suggest there might effectmeasuremodificationon:Themultiplicativescale?There appears to be effect measure modification on the multiplicativescale. The CIR among men who are older than 55 years is nearly doubletheCIRamongmenwhoare55yearsoryounger.Theadditivescale?It seems like there is effect measure modification on the additive scale.The CID among men who are plder 55 years is more than double the CIDamongmenwhoare55yearsoryounger.d.For the stratified data, perform the test of homogeneity for thecumulative incidence ratio. State the null and alternativehypotheses. Interpret your findings. H0: There is no effect measure modification of the association betweenever versus never smoking and oral or bladder cancer across the strataof age on the multiplicative scale.HA: There is effect measure modification of the association between everversus never smoking and oral or bladder cancer across the strata of ageon the multiplicative scale.H=i=1I[ln(^CIRi)ln^CIRMH]2^Vari[ln(^CIRi)]xI1212
CIR1.833.56Var(ln(CIR)0.06210.0679Ln(CIR)0.60431.2697^CIRMH=aiN0iTibiN1iTi=93×11044057+72×1240263319×29534057+18×13932633=25.31+33.9113.83+9.52=59.2223.35=2.536ln(CIRMH)=0.9306H=i=1I[ln(^CIRi)ln^CIRMH]2^Vari[ln(^CIRi)]=(0.60430.9306)20.0621+(1.26970.9306)20.0679=1.7145+1.6935=3.408x2with1Pr[x12>3.41]=0.065At the 0.05 level of significance, we fail to reject the null hypothesis. We do

#### You've reached the end of your free preview.

Want to read all 6 pages?

• Summer '14
• FrancisCook