In: Statistics and Probability
2. There is a large area of Bangladesh where the groundwater is contaminated by arsenic. Residents of this area suffer from elevated rates of cancer, and of pre-cancerous skin lesions, because they often have no alternative but to drink the arsenic-contaminated water. Many efforts are underway to solve this problem. In one study, investigators looked for evidence that the effect of arsenic on pre-cancerous skin lesions might be increased or decreased by other factors – effect modifiers. One such possible factor is pesticide use which is common in this agricultural region. The table shows the results of a 6-year closed prospective cohort study in which residents were followed and periodically examined for the occurrence of pre-cancerous skin lesions. All these people are exposed to arsenic in drinking water, and so this table does not show that exposure. These data divide the cohort into 3 groups based on how long they were exposed to pesticides. a. b. c. d. a. What measure of disease frequency can be calculated in these data? b. Calculate this measure for each of the 3 exposure groups. c. For the group with no pesticide exposure, calculate a 95% confidence interval for the measure of disease frequency. What does this interval mean? d. Calculate a ratio measure of association, comparing the risk of skin lesions in those with less than 7 years of exposure to those with no exposure, and another measure of association comparing those with 7 and over years of exposure to those with no exposure. Interpret these 2 measures of association. e. Is there evidence that pesticide exposure affects the risk of skin lesions in this population? Explain.
solution:
A) We can calculate the Incidence rate/Proportion of the occurrence of lesion using this data.
Incidence rate : It is the fraction of individuals exposed out of the total observed individuals. It can be expressed as a percentage as well.
B) Not Exposed:
Incidence rate = #Skin Lesions/Total followed
= 222/1924 = 0.115 or 11.5%
Exposed(0-7):
Incidence rate = #Skin Lesions/Total followed
= 138/939 = 0.147 or 14.7%
Exposed(7+):
Incidence rate = #Skin Lesions/Total followed
= 250/1114 = 0.224 or 22.4%
C).
p=222/1924=0.115385
CI=p+/-Za/2*p(1-p)/n
CI=0.115385+/-1.96*0.115385(1-0.115385)/1924
CI=0.115385+/-0.014
CI=0.1011 to 0.1297
We can say we 95% confidence that the true mean lies between 0.1011 to 0.1297
D)
RR (less than 7 yrs to 0 yrs)=138/939/222/1924=1.2736 (Those with less than 7 yrs of pesticide exposure have 1.27 times more risk of developing skin lesions as compared to 0 yrs of pesticide exposure)
RR (7 and over yrs to 0 yrs)=250/1114/222/1924=1.9449 (Those with 7 yrs and over of pesticide exposure have 1.94 times more risk of developing skin lesions as compared to 0 yrs of pesticide exposure)
E)
Looking at the results above (d), we can see that risk of skin lesions increase as pesticide exposure increases and hence we conclude that pesticide exposure affects the risk of skin lesions