•   Please only “cut-and-paste” relevant SPSS output for each question
•   You can upload Word or pdf file.
•   The due date for this Exam is Sunday April 1, 2018 after which Blackboard will refuse submission

1 A university medical center urology group was interested in the association between Prostate-Specific Antigen (PSA) and a number of prognostic clinical measurements in men with advanced prostate cancer. Data were collected on 97 men who were about to undergo radical prostectomies. The other measures are Cancer Volume, Prostate Weight, Patient Age, the amount of Benign Prostatic Hyperplasia, presence (1) or absence (0) of Seminal Vesicle Invasion, Capsular Penetration and Gleason Score.

The goal here is to develop a “good” model to predict PSA based on the other prognostic clinical measurements using a regression model. I mention a few pointers below, but these are just suggestions and not specific guidelines. You should try different models and different things and put your findings in a report including interpretation and usefulness of the statistical analysis. You also may have to go back and forth among some of the following.

•   Scatterplot for pairs of variables (PSA versus another)
•   Correlation among pairs of variables
•   Multiple Regression
•   Association between PSA and individual covariates
•   How good is the fit?
•   A “good” model selection via stepwise method(s).
•   Regression assumptions – are they satisfied?
•   If some of the assumptions are of concern, try LOG or SQRT transformation of the response (Log(PSA)or SQRT(PSA) ). Check the assumptions after refitting the model.
•   Interactions
•   Multicollinearity

Exam 2 Page 1

2 (Text, p 341) Radioactive radon is an inert gas than can migrate from soil and rock and accumulate in enclosed areas such as underground mines and homes. The radioactive decay of trace amounts of uranium in Earth’s crust through radium is the source of radon, or more precisely, the isotope radon-222. Radon-222 emits alpha particles; when inhaled, alpha particles rapidly diffuse across the alveolar membrane of the lung and are transported by the blood to all parts of the body. Due to the relatively high flow rate of blood in bone marrow, this may be a biologically plausible mechanism for the development of leukemia. The data here come from a case-control study to investigate the association between indoor residential radon exposure and risk of childhood acute myeloid leukemia. The variables are

– Disease (1, case, 0, control)
– Radon (radon concentration in Bq/m3)
– Some characteristics of the child: gender (1, male, 2, female), race (1, white; 2, black; 3, Hispanic; 4, Asian; 5, others), Down’s syndrome (a known risk factor for leukemia; 1, no; 2, yes)
– Risk factor from the parents: Msmoke (1, mother a current smoker; 2, no, 0, unknown), Mdrink (1, mother a current alcohol drinker; 2, no; 0, unknown), Fsmoke and Fdrink.

GCC 547 Principles of Biostatistics II

The response or the variable of interest is Disease status which is binary. The goal here is to develop a “good” model to predict disease status based on the other variables. There are again no exact guidelines. I mention some specific questions below. You can try different models and different things and put your findings in a report including interpretation and usefulness of the statistical analysis.

•   What are your findings concerning the adjusted contribution of each factor (covariate) in the multiple regression model?
•   Do the smoking-drinking variables (either parent) relate significantly to the disease of the child?

3. (Text, p 428) Pneumocystis carinii pneumonia (PCP) is the most common opportunistic infection in – HIV–infected patients and a life-threatening disease. PCP is a consideration factor in mortality, morbidity and expense and recurrences are common. The dataset that we consider here have

– Treatments, coded as A and B

– Patient characteristics: baseline CD4 count, Gender (1, male; 0, female), race (1, white; 2, black; 3, other), Weight (lb), homosexuality (1, yes; 2, no)

– PCP recurrence indicator (1, yes; 0, no), PDATE or time to recurrence in months

– DIE or survival indicator (1, yes; 0, no), DDATE or time to death (or, date last seen for survivors) in months

Consider each of these “time-to-event” endpoints:
Exam 2 Page 2

GCC 547- Principles of Biostatistics II Dr. Sanjib Basu Spring 2018

Exam 2

Page 3

Relapse (PDATE) Death (DDATE)

Death or relapse (whichever comes first).
Create a new variable as MIN(DDATE, PDATE) using Transform -> Compute in SPSS.

For each of these endpoints:

1. Estimate the survival function for each treatment. Compare the two treatments?
2. Compare the survival functions of men and women.The next question is related to proportional hazard modeling
3. Comment on the association of the other variables with the time-to-event outcomes?

GCC 547- Principles of Biostatistics II` Dr. Sanjib Basu Spring 2018