Analysis

Survey123

Analysis

Don't you love my presentation style?

Assignment Aim

Apply concepts learned in Class (and Other Classes)

• Assess your ability to navigate Survey123
• Assess your ability to apply Epi methods
• Assess your ability to review datasets, assess missing data and how that might impact an analytic plan, and ensure variables were in appropriate format for analysis
• Apply concepts and examples discussed in prior lectures
• Practice applied epidemiology in a real-world scenario
• Applied epidemiologists often have to make decisions with limited information, using messy data, using analytic framework and outbreak investigation steps as a guide

How many individuals were defined as cases and how many controls were recruited?

Q1

Need to navigate to the responses to "What is the case status?"

Produce a table, one for ill individuals and one for non-ill individuals that summarizes basic demographics: age and gender. Do you notice any differences?

Q2

Need to filter by the question "Was the patient ill?"

Q3

Were any controls ill? Do they meet the case definition?

Double Filter

Was the patient ill? -yesWhat is the case status? -control Fit case definition? This is a chance to show your understanding of a case definition, do these controls meet the criteria for symptoms and lab confirmation?

Q4

Using Excel, produce an epidemic curve visualization. What sort of outbreak does this appear to be based on your epidemic curve?

But then Survey123's Analytics....

Q5

Produce a table describing frequencies and attack rates of variables you think are key to this investigation.

• Frequency and Attack Rates for variables of interest
• Should lead you to a hypothesis of possible exposure

Q6

Produce a table calculating the OR of the variables you think are key to this investigation.

• Odds ratios for variables of interest
• Should lead you to a hypothesis of possible exposure

Q7

What variables would you include in a multivariable regression? What type of regression would you perform?

• Need to be able to explain your rationale for your chohice of variables to include in the multivariable regression analysis
• ORs>1 - Should make sense based on what you did in Q5 & Q6
• Any other reasons?
• Logistic regression
• Why? Binary outcome

Q8

Write a one-paragraph summary of your findings.

• Summary of your findings.
• Summarizes descriptive data
• What variables are associated with the outcome (illness, being a case)?
• Why?
• ORs>4 implies strong relationship
• 4<ORs>2 implies moderate association

Q9

What next steps and/or control measures would you recommend based on your findings?

• Communication with the vendors
• Communication with the public

These should be based on your findings.

Supplementary Resources

Questions?