Bio Statistics
Homework
Cancer, Endocrinology
Obesity is very common in American society and is a risk factor for breast cancer in postmenopausal women. One mechanism explaining why obesity is a risk factor is that it may raise estrogen levels in women. In particular, one biomarker of estrogen, serum estradiol, is a strong risk factor for breast cancer. To better assess these relationships, researchers studied a group of 151 African American and 60 Caucasian premenopausal women. Adiposity was quantified by two different measures: BMI = weight (kg)/height2 (m2) and waisthip ratio (WHR) = waist circumference/hip circumference. BMI is a measure of overall adiposity, whereas WHR is a measure of abdominal adiposity. In addition, a complete hormonal profile was obtained, including serum estradiol (ES_1). Finally, other breast-cancer risk factors were also assessed among these women, including (1) ethnicity (ETHNIC = 1 if African American, = 0 if Caucasian), (2) age (ENTAGE), (3) parity (NUMCHILD = number of children), (4) age at first birth (AGEFBO), (5) any children (ANYKIDS = 1 if yes, = 0 if no), (6) age at menarche (AGEMNRCH = age when menstrual periods began). The data are provided in Data Set ESTRADL. DAT at www.cengagebrain.com
10 Is there a crude association between either measure of adiposity (BMI, WHR), considered separately, and serum estradiol?
11 Are these relationships similar for Caucasian and African American women?
12 Do the relationships between the adiposity measures and serum estradiol persist after controlling for the other breast-cancer risk factors in list items 1 to 6?
13 It is well known that African American women have higher levels of obesity than Caucasian women. Are there differences between estradiol levels for African American women and Caucasian women after controlling for obesity?
Regression
14 Is there evidence of nonlinearity for age or for height?
Genetics, Diabetes
Suppose we have separately analyzed the effects of 10 SNPs comparing people with type I diabetes vs. controls. The p-values from these separate analyses are given in
Table 12.44
.
Table 12.44
1 Use the Bonferroni method to correct for multiple comparisons. Which SNPs show statistically significant effects?
2 Use the FDR method to correct for multiple comparisons using an FDR = .05. Which SNPs show statistically significant effects? How do the results compare with those in Problem 1?
Hypertension
A two-way ANOVA was run using PROC GLM of SAS comparing mean diastolic blood pressure (DBP) by study group and sex. The results are given in
Table 12.41
.
Table 12.41
3 Summarize the findings in a few sentences.
Ophthalmology
The term retinitis pigmentosa (RP) refers to a group of hereditary, retinal pigmentary degenerations in which patients report night blindness and loss of visual field, usually between the ages of 10 and 40 years. Some patients lose all useful vision (i.e., become legally blind) by the age of 30 years, while others retain central vision even beyond the age of 60 years. A specific gene has been linked to some types of RP where the mode of genetic transmission is autosomal dominant. The most reliable methods of following the course of RP in humans is by using the electroretinogram (ERG), which is a measure of the electrical activity in the retina. As the disease progresses, the patient’s ERG amplitude declines. The ERG amplitude has been strongly related to the patient’s ability to perform routine activities, such as driving or walking unaided, especially at night. One hypothesis is that direct exposure of the retina to sunlight is harmful to RP patients, so many patients wear sunglasses. To test the sunlight hypothesis, researchers introduced this gene into a group of mice and mated them over many generations to produce a group of “RP mice.” Then they randomly assigned the mice to lighting conditions from birth that were either (1) light, (2) dim, or (3) dark. A control group of normal mice was also randomized to similar lighting conditions. The mice had their ERG amplitudes (labeled BAMP and AAMP for B-wave amplitude and A-wave amplitude, respectively), which correspond to different frequencies of light, measured at 15, 20, and 35 days of life. In addition, the same protocol was used for a group of normal mice except that only BAMP was measured. The data for both RP mice and normal mice are available in the Data Set MICE.DAT and the documentation in MICE.DOC (both at www.cengagebrain.com).
4 Analyze the data regarding the sunlight hypothesis, and summarize your findings. (Hint: Estimate a slope for each ERG amplitude for each lighting-condition group, and compare the slopes among the light, dim, and dark groups using either ANOVA or regression methods. Do separate analyses for AAMP and BAMP. Consider appropriate data transformations to ensure approximate normality for the outcome measures.)