Appropriately labelled rows and columns

CAM625 2018 Module 4 Assignment

Instructions

The stated maximum number of words for individual questions are a guide, not a target. Marks will be deducted for overly verbose answers; please be brief, you just need to demonstrate that you have understood the question and answer only what you are asked to do. There is no need to do extra work.

Please do not discuss (neither personally nor on the discussion board) specific findings or results from this assignment. We anticipate that you may have some queries about R code, please try to phrase your questions carefully when posting to the discussion board, or to be safe just email us privately.

Your task is to prepare a report of the data and models. Firstly showing summary data of selected variables. Then the association between gestation length and being overweight, and finally the association between parity of three or more live births and being overweight.

You are required to present neat tables we suggest you use Microsoft Word to create these, please ask for help for this task if you need it.

Data cleaning

Download the gestation dataset (assign4data.csv file) from the Module contents. Copy the following code into an R script file, modify the directory path. This code will create the variables that you need to use for your analysis.

2

Task 1

For your table use Stannus et al (2015) or Thompson et al (2017) as a guide. Examine Table 1 of Stannus et al (2015) or Thompson et al (2017) to see how we wish you to present these summary data: Stannus shows two columns one for low Leptin and one for high Leptin, Thompson shows two columns one for skin cast grades 3/4 and one for grades 5/6. Your table will have one column for participants with normal bmi (not overweight) at follow-up and one for participants with classified as overweight at follow-up. The table should not simply be a cut and paste of R output. You are required to present the table in your Word document as you would for publication in a journal, including a table title and caption (if required), appropriately labelled rows and columns, and data rounded to a meaningful number of decimal places.

3. Write a brief summary paragraph of the results of the tests above, giving basic interpretations of the differences between the two groups that you observe.

• overweight (ow),
• mother’s age (age),
• mother’s race (race),
• parity of 3 or more (par_3),

2. Report the regression/beta coefficients and their 95% confidence intervals in a table (this table will be combined with the one from Task 3 below so only prepare one table for Tasks 2 and 3). An example can be found can be found in Stannus 2015 (Table 2). Again prepare the table with a title, captions, labelled rows and columns and appropriately rounded numbers.

(maximum 300 words)
In epidemiology/biostatistics we are usually concerned with describing the true association between a key exposure and an outcome. To estimate the “true” association we wish to adjust our models for any confounders by adding them to the model to create a multivarible model. The key exposure in this case is overweight and the outcome is gestation length. In this task you will add all of the variables from Task 2 into the same model.

1. Fit the multivariable model including all the variables in Task 2.

4

Task 4

3. Report the effect size (this is not the regression coefficients) and their 95% confidence intervals in a table (this table will be combined with the one from Task 5 below so only prepare one table for Tasks 4 and 5).

An example of the table can be found can be found in Thompson et al 2017 (Table 2), be aware that Table 2 in Thompson et al contains four columns because they examine three adjusted models, you will only need two columns (unadjusted and adjusted).

1. Fit a multivariable model, include all of the variables in Task 4.

2. Add to your table from Task 4 the effect sizes (and 95% confidence intervals) from a multivariable model in a new column. Line up the rows as in Table 2 of Thompson et al (2017).