Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

ST332 & ST409 Medical Statistics: 2023-2024

Assignment [20%]

Deadline: 13:00 Friday 20th April 2024

Background

Clinicians at University Hospitals Coventry & Warwickshire (UHCW) NHS Trust have studied  whether elevated levels (above 2.9 ng/mL) of a biomarker, carcinoembryonic antigen (CEA), after surgery (typically collected within 2 weeks following surgery) for patients undergoing with colorectal cancer are prognostic for all-cause mortality. They collected data on co - morbidities diagnosis of cardiovascular disease (CVD) prior to surgery, diagnosis of Chronic obstructive pulmonary disease (COPD) prior to surgery, in addition to smoking status which   is thought to affect CEA levels and age at the time of surgery.

Dataset

The R dataset crcsurg.RData contains synthetic data on 160 patients treated at UHCW NHS Trust over 15 years with a median follow-up of 5.7 years. The dataset only considers death as an outcome and contains the following variables and associated codes;

id

(Patient ID number)

CEA

(0=normal,1=elevated)

Smoker

(0=no,1=yes at time of surgery)

Male

(0=female,1=male)

Age

(age in years at time of surgery)

CVD

(0=none,1=CVD at time of surgery)

COPD

(0=none,1=COPD at time of surgery)

pys

       (person-years follow-up since surgery)

dead

(0=alive,1=dead)

Assignment Tasks

PART A [Maximum 5 pages – 15 marks]

Answer the following questions based on the crcsurg.RData dataset described above. Note it is not necessary to write a report – just answer the questions,  but do not include R code or paste R output – marks will be lost for this.

1.   Undertake an Exploratory Data Analysis of the crcsurg.RData dataset, in particular identify any covariates which may be potential confounders. [5 marks]

2.   Produce a summary table of the numbers of deaths and person-years data, based on the CEA level of patients. Calculate and interpret the associated Incidence Rate Ratio (IRR) and it’s 95% CI and P-value. [3 marks]

3.   Present and interpret an appropriate log-linear Poisson regression model that captures the effect of elevated CEA levels after adjusting for other relevant variables.

Carefully describe the model (defining any variables you refer to), how your model was chosen, how the results compare to those in Question 2, and any interesting/pertinent features. [5 marks]

4.   Your clinical colleagues suggest that a 10% relative increase in mortality due to elevated CEA levels would be clinically important to detect. Describe how you would design a future study based on this information (and the results of your analyses in questions  1 - 3). [2 marks]

PART B [500 words maximum - 5 marks]

Based on your answers to  PART A write a Press Release about the study for the University to make available to the  media.

Marks will be awarded for;

.   Appropriate interpretation of the results in PART A

.   Discussion of strengths and limitations of the study

.   Use of  non-technical language for an informed lay audience

.   Inclusion of an “anonymous” quote from you as  Researcher <Joe Bloggs>”

.   Discussion of potential next steps/further research

You can find examples of Press Releases on the University website at;

https://warwick.ac.uk/newsandevents/pressreleases/and two good recent health

-related examples are;

https://warwick.ac.uk/newsandevents/pressreleases/?newsItem=8a17841b8defef9 8018df5

c098261254and

https://warwick.ac.uk/newsandevents/pressreleases/?newsItem=8a17841a8d7973 0b018d9 e2bbb0e054b

General Instructions

It is expected that PART A and B will be professionally presented (e.g. using LaTeX, WORD or R Markdown) and submitted on Moodle as a single pdf document. Do NOT put your name on the document or in the filename, please just use your student number. For PART B include a word count at the end of the Press Release. No appendices are allowed.

General questions about the assignment can be posted (anonymously if you wish) to the Discussion Forum on Moodle page for ST332, and these will be answered, questions emailed directly will not be answered.