Biostatistics POPH90013 – 2023 – SM2 – Assignment 4
Due: Tuesday 31th Oct, 11.59 pm
The maximum mark for this assignment is 80. It forms 40% of the final grade for this subject.
Your assignment should be submitted via Gradescope as a PDF document.
Please put your student ID number in the header of the document.
➢ Please do not copy/paste any Stata output in your assignment document. Instead,
format any results you want to show in a way that would be suitable for inclusion in
a report or journal article.
➢ For questions where you are asked to calculate the answers by hand, please show
your workings.
➢ For all questions that state “Using Stata...”, please write the Stata command in the
beginning of that question as marks are allocated per question for the Stata
command.
➢ Please interpret all estimates as shown in Stata practical solution document.
For this assessment, you will need to download and open the files
“Assignment4_wound.dta” and “Assignment4_pain.dta” from Canvas. Also, we
recommend that you perform all Stata tasks via a do-file (you can use the do-files from the
Stata practicals located on Canvas, as a guide).
Section A - Wound Healing Dataset
The Wound Healing Society defines a chronic wound as one that has failed to proceed through
an orderly and timely reparative process to produce anatomic and functional integrity within
an expected period. Chronic wounds represent a significant annual burden on the Australian
health care system, with direct health care costs reaching $2.85 billion. Several factors can
interfere with one or more phases of the wound healing process, thus causing improper or
impaired wound healing. Such factors include age, stress, diabetes, obesity, medications,
alcoholism, smoking, and nutrition. A better understanding of the influence of these factors
on repair may lead to therapeutics that improve wound healing and resolve impaired wounds.
For this section of the assignment, you will be using the dataset “Assignment4_wound.dta”,
a prospective cohort study of 750 wound patients investigating the risk factors for wound
infection, where the patients were given uniform treatment over 12 weeks.
Table 1: Description of the variables in the Assignment4_wound.dta dataset
Variable name Description
id Study participant identification number
age Age (years)
sex_male Sex (0 – Female, 1 – Male)
bmi Body mass index (kg/m2)
smoke Smoking status (0 – Non-smoker, 1 – Smoker)
alc Alcohol consumption per week (ml/week)
stress Stress score (units, Range: 0 - No stress, 10 - Maximum stress)
diab Type II diabetes (0 – No, 1 – Yes)
infect Was the wound infected at any time in twelve weeks? (0 – No, 1 – Yes)
*All other variables except wound infection (infect) were measured at hospital admission.
Question 1 [21 marks]
Some common risk factors for wound infection are age, sex, stress, diabetes, obesity, high
alcohol consumption and smoking.
a) Provide a table (in a format that would be acceptable for a report or journal article) that
summarises the distribution of the outcome of interest wound infection and the potential risk
factors, age (years), sex, stress (units), diabetes, body mass index (BMI, kg/m2), alcohol
consumption (ml/week) and smoking. Describe the study population using this table. [6
marks]
One of the research questions of this study was to explore if diabetic people were more
susceptible to wound infection. From related literature the study investigators were able to
further identify that a patient’s sex may confound the association between diabetes and
wound infection. The directed acyclic diagram (DAG) is shown below.
b) Calculate using Stata and interpret, the unadjusted odds ratio for the association between
diabetes (diab) and wound infection (infect). Write down the Stata command you used
to obtain this odds ratio as well. [2 marks]
c) Calculate using Stata, and interpret the adjusted odds ratio, corresponding 95% confidence
interval for the population odds ratio and the two-sided p-value for the association between
diabetes (diab) and wound infection (infect), adjusted for sex (sex_male). Write down
the Stata command you used to obtain these results as well. [4 marks]
Diabetes
Wound
Infection
Sex
d) Investigate the validity of your findings by comparing the unadjusted and adjusted odds ratios
calculated in Questions 1b and 1c. And explore the associations between the exposure and
confounder, and outcome and confounder to explain your observations. [5 marks]
e) A group of investigators want to design a randomised controlled trial to explore the effect of
a diabetes-intervention drug on wound infection. Wound patients who have diabetes will be
randomly assigned in a 1:1 ratio to receive either the new diabetes drug (i.e., the intervention
group) or continue their usual diabetes medication (i.e., the control group). The investigators
have estimated the proportion of wound infections among participants continuing their usual
diabetes medication to be similar to the proportion of wound infections among diabetics in
the current study population. They are interested in detecting an absolute reduction of 20%
in the proportion of wound infections among participants assigned to the new diabetic drug.
Using Stata, calculate the sample size required for this new study assuming a significance level
of 3% and a power of 80%. Write down the Stata command you used to calculate this sample
size as well. [4 marks]
Question 2 [14 marks]
The study investigators were interested in analysing the cross-sectional hospital admission
data on smoking status (smoke) and stress scores (stress) and exploring whether patients
who smoked had increased stress levels. From related literature the study investigators were
able to further identify that a patient’s age (age) may confound the association between
Smoking status and stress. The directed acyclic diagram (DAG) is shown below.
Smoking status Stress
Age
a. Provide a graph visualising the association between stress and smoking status and
comment on the association. Write down the Stata command you used to obtain this. [3
marks]
b. Using Stata, fit a linear regression model to obtain an estimate for the unadjusted
association between stress and smoking status, and provide an interpretation for this
estimate. Write down the Stata command you used to obtain this. [2 marks]
Participants with an age of at least 45 years were considered by researchers to be old. Using
the age variable, generate a new binary variable named age_bin as; participants with an
age less than 45 [coded as 0, “young”] and participants with an age of at least 45 years [coded
as 1, “Old”]. Use this new binary variable age_bin for age in Questions 2c, 2d and 2e.
Write down the Stata commands used.
c. Using Stata, obtain the frequency and proportion of patients who are old in this study.
Write down the Stata command you used to obtain this proportion as well. [3 marks]
d. Using Stata, fit a linear regression model to obtain the adjusted association between the
outcome stress score, exposure smoking status and confounder Age. Interpret the
adjusted association for stress and smoking status using the estimate, corresponding 95%
confidence interval and p-value. Write down the Stata command you used to obtain these
results as well. [5 marks]
e. Comment on whether the association between stress and smoking status is confounded
by Age in this study, by comparing the unadjusted and the adjusted estimates calculated
in Questions 2b and 2d. [1 mark]
Question 3 [10 marks]
A researcher would like to use this cross-sectional hospital admission data to investigate
whether individuals with wound infection were at a higher risk of being smokers.
a.) Using Stata, obtain the frequency and proportion of individuals with wound infections in
12 weeks (infect), separately for smokers and non-smokers (smoke). Write down the
Stata command you used to obtain these results as well. [2 marks]
b.) Calculate using Stata and interpret, the risk difference, corresponding 95% confidence
interval for the population risk difference, and a two-sided p-value, for the unadjusted
association between smoking status (smoke) and wound infections in 12 weeks
(infect). Write down the Stata command you used to obtain these results as well. [3
marks]
c.) Using the findings of this study, the researcher decides to design a cross-sectional study
to determine the population proportion (also known as the prevalence) of individuals with
wound infection. Calculate by hand, the sample size required to obtain an estimate of this
population prevalence of people with wound infection with a precision of ±3.5% for the
95% confidence interval. [5 marks]
Section B - Pain Dataset
More than 500 million people are injured each year, with a large proportion of them
sustaining an orthopaedic injury which requires surgery. Post-surgery pain may lead to
prolonged recovery time, a severe adverse impact on a patient’s health, dissatisfaction with
health care and if not controlled adequately, may persist to become a chronic pain condition.
For this section of the assignment, you will be using the dataset “Assignment4_pain.dta”, a
random sample of 200 patients from an observational cohort study conducted at the Royal
Melbourne Hospital. The aim of this study was to explore risk factors of severe acute pain in
the post-anaesthesia care unit (PACU).
Table 1: Description of the variables in the Assignment4_pain.dta dataset
Variable name Description
id Study participant identification number
sex_female Sex (0 – Male, 1 – Female)
age Age (years)
height Height (cm)
bmi Body Mass Index (BMI, kg/m2)
anaesdur Anaesthesia duration (min)
opsite Surgery site (0 – Other sites, 1 – Lower leg, ankle, foot)
pain Severe acute pain in the post-anaesthesia care unit (0 – No, 1 – Yes)
Question 4 [20 marks]
One of the research questions of this study was to explore if participants having surgery on
the lower leg, ankle or foot were more susceptible to severe acute pain in the post-
anaesthesia care unit (PACU). From related literature the study investigators were able to
further identify that a patient’s sex may confound the association between surgery site and
severe acute pain in the PACU. The directed acyclic diagram (DAG) is shown below.
a. Visualise the unadjusted association between surgery site (opsite) and severe acute
pain in the PACU (pain) by completing the 2×2 table below. [2 marks]
Surgery Site
Severe acute pain in the PACU
Total
Yes No
Lower leg, ankle, foot
(Group 1)
Other sites (Group 0)
Total
b. Calculate by hand, and interpret the unadjusted odds ratio for the association
between surgery site and severe acute pain in the PACU. [3 marks]
c. Visualise the association between surgery site (opsite) and severe acute pain in the
PACU (pain), stratified by sex (sex_male) by completing the table below. [4 marks]
Surgery Site
Female Male
Severe acute pain in the PACU Severe acute pain in the PACU
Yes No Yes No
Surgery site
Severe acute
pain in the
PACU
Sex
Lower leg, ankle, foot
(Group 1)
Other sites (Group 0)
Total
d. Calculate by hand and interpret the odds ratio for the association between surgery
site and severe acute pain in the PCAU separately for males and females. How do the
stratum-specific odds ratios compare with each other, and the unadjusted odds ratio
calculated in Question 4b? [4 marks]
e. Calculate by hand and interpret the Mantel-Haenszel estimate of the pooled odds
ratio for the association between surgery site and severe acute pain in the PCAU, adjusted
for sex. [3 marks]
f. Comment on any confounding observed by considering any changes between the
unadjusted and the adjusted odds ratios calculated in Questions 4b and 4e. Explore the
associations (by hand or using Stata) between the exposure and confounder, and outcome
and confounder to explain your observations. [4 marks]
Question 5 [ 15 marks]
A secondary aim of this study was to investigate the factors influencing Anaesthesia
duration (anaesdur). The researchers believed that some factors influencing anaesthesia
duration are age, sex, BMI, height, and surgery site.
a. Provide a table (in a format that would be acceptable for a report or journal article) that
summarises the distribution of the outcome of interest anaesthesia duration
(anaesdur) and the potential exposures, age (age), sex (sex_female), BMI (bmi),
and surgery site (opsite)). Describe the study population using this table. [6 marks]
b. Provide a graph visualising the association between anaesthesia duration and sex, and
comment on the association. [3 marks]
c. Provide a graph visualising the association between anaesthesia duration and BMI, and
comment on the association. [3 marks]
d. Provide a graph visualising the association between anaesthesia duration and surgery
site, and comment on the association. [3 marks]