screening coverage in the past month

[x] What is the coverage of mid-upper arm circumference (MUAC) screening among the children in the sample?
[x] What is the coverage of mid-upper arm circumference (MUAC) screening among the children in the sample by sex?
[x] What is the coverage of mid-upper arm circumference (MUAC) screening among the children in the sample by location?
[x] What is the coverage of nutritional oedema screening among children in the sample?
[x] What is the coverage of nutritional oedema screening among children in the sample by sex?
[x] What is the coverage of nutritional oedema screening among children in the sample by location?

@OxfordIHTM/charmander

@OxfordIHTM/charmander here are some notes on answers to questions that your team and other teams have asked in today's session:

As pointed out by your teammate @quyentrano, the study round stratification/grouping is just as important. Reporting results for the entire dataset doesn't make any sense as about half of the entire dataset was collected at a different dates and times as the other half. It makes a lot more sense for the results/answers for your problem to be shown/provided by Baseline and Endline survey rounds. Well spotted @quyentrano!
With regard to the indicator on screening coverage for the past month, it is worded this way because the responses shown in the variable muac_screening and oedema_screening is to the question "Has your child been measured like this (MUAC or oedema) in the past month?". Hence the response in these variables already pertain to recall of mother over the past month.
Important to note that MUAC and oedema screening is performed as early as 0 months old up to 59 months old.
Here again is a recap of the variables of the nut dataset:

variable name	variable description
survey_round	Survey round; either Baseline or Endline
survey_data	Date interview was conducted/data collected was performed
county	County location respondent comes from; either Grand Bassa or Urban Montserrado
sex	Sex of child; 1 = Male; 2 = Female
age	Age of child in whole months
muac	Mid-upper arm circumference (MUAC) measurement in centimetres
oedema	Presence or absence of nutritional oedema; 1 = Yes; 2 = No
muac_screen	In the past month, has the child's MUAC been measured? 1 = Yes; 2 = No; 99 = no answer/cannot remember
oedema_screen	In the past month, has the child been checked for nutritional oedema? 1 = Yes; 2 = No; 99 = no answer/cannot remember
cov_status	Is the child currently receiving treatment with peanut butter medicine/Plumpynut? 1 = Yes; 2 = No

Although this is not part of the problem set that you are working with (i.e., screening coverage), it might end up helpful later on for additional analysis the case definition for the different forms of acute malnutrition:
- global acute malnutrition (GAM): MUAC < 12.5 cms or positive for nutritional oedema for children 6-59 months old
- moderate acute malnutrition (MAM): MUAC < 12.5 and MUAC >= 11.5 cms and negative for nutritional oedema for children 6-59 months old
- severe acute malnutrition (SAM): MUAC < 11.5 cms or positive for nutritional oedema for children 6-59 months old
@alaa-dafallah asked a question during the session on whether we need to count the total number of rows of data to be able to answer the question on screening coverage. In theory, yes. What is important for team @OxfordIHTM/charmander to conceptualise is the formula/equation to calculate screening coverage. Any coverage indicator is by definition a proportion indicator, i.e. number of the target population receiving a specific service or benefit divided by the total number of the target population. This can be expressed like this:

$$ \text{coverage} ~ = ~ \frac{\text{Number of target population receiving a specific service or benefit}}{\text{Total number of target population}} $$

So, important to define the following:

- target population (this may mean what is the criteria to be eligible for the service/benefit);
- target population that is receiving the specific service/benefit

Once you are able to define these, then apply these definitions to the data using your data wrangling skills. There are many ways to do this but mistakes arise when the definitions are not clearly specified to begin with.

Team @OxfordIHTM/charmander, I hope you are all doing well.

I haven't seen any pull request from the team and the branches from team members that I have looked at have not shown any new contributions. So, just want to check if everything is ok and if you guys need help? No pressure. There is no expectation to contribute but I hope if indeed you have something to share or something that you are working on, would be good to see it so that I can provide feedback.

For our last session next week, get your R script completed and then start on preparing your section in the hackathon report. In the project/repository, you would have noticed an .Rmd file named coverage_assessment_report.Rmd. When you open that file, you should see a pre-prepared report format for the results of this hackathon and you specific section outlined in the report.

Using what you have learned in our R Markdown session, start adding text and paragraphs for your section of the report to present the results of your analysis. Please use tables to present the counts asked in the challenge and then think of how you can show your results creatively using plots.

I will walk around/move around and help you get started during our last session.

Team @OxfordIHTM/charmander, a table that looks like this on the report will be a good summary of your analysis results:

Urban Montserrado

MUAC screening

	Baseline	Endline
Male	Number of males screened by MUAC at baseline (% of males screened by MUAC out of total males at baseline)	Number of males screened by MUAC at endline (% of males screened by MUAC out of total males at endline)
Female	Number of females screened by MUAC at baseline (% of females screended by MUAC out of total females at baseline)	Number of females screened by MUAC at endline (% of females screened by MUAC out of total females at endline)
Total	Total number screened by MUAC at baseline (% of total screened by MUAC at baseline)	Total number screened by MUAC at endline (% of total screened by MUAC at endline)

Oedema screening

	Baseline	Endline
Male	Number of males screened for nutritional oedema at baseline (% of males screened for nutritional oedema out of total males at baseline)	Number of males screened for nutritional oedema at endline (% of males screened for nutritional oedema out of total males at endline)
Female	Number of females screened for nutritional oedema at baseline (% of females screended for nutritional oedema out of total females at baseline)	Number of females screened for nutritional oedema at endline (% of females screened for nutritional oedema out of total females at endline)
Total	Total number screened for nutritional oedema at baseline (% of total screened for nutritional oedema at baseline)	Total number screened for nutritional oedema at endline (% of total screened for nutritional oedema at endline)

Grand Bassa

MUAC screening

	Baseline	Endline
Male	Number of males screened by MUAC at baseline (% of males screened by MUAC out of total males at baseline)	Number of males screened by MUAC at endline (% of males screened by MUAC out of total males at endline)
Female	Number of females screened by MUAC at baseline (% of females screended by MUAC out of total females at baseline)	Number of females screened by MUAC at endline (% of females screened by MUAC out of total females at endline)
Total	Total number screened by MUAC at baseline (% of total screened by MUAC at baseline)	Total number screened by MUAC at endline (% of total screened by MUAC at endline)

Oedema screening

	Baseline	Endline
Male	Number of males screened for nutritional oedema at baseline (% of males screened for nutritional oedema out of total males at baseline)	Number of males screened for nutritional oedema at endline (% of males screened for nutritional oedema out of total males at endline)
Female	Number of females screened for nutritional oedema at baseline (% of females screended for nutritional oedema out of total females at baseline)	Number of females screened for nutritional oedema at endline (% of females screened for nutritional oedema out of total females at endline)
Total	Total number screened for nutritional oedema at baseline (% of total screened for nutritional oedema at baseline)	Total number screened for nutritional oedema at endline (% of total screened for nutritional oedema at endline)

Team @OxfordIHTM/charmander, I have reviewed the only pull request that came from your group via @quyentrano.

I have given feedback via GitHub which @quyentrano would see but I will tag all of you as well so you can see my feedback.

I have merged this to the main branch myself and then have made some edits based on the feedback I have given.

I think you may not have had the time to add literate code in the R Markdown so I used the R script provided to add code and text to the R Markdown.

I will be reviewing each of your individual branches and give feedback directly to each of you individually.

Well done!

Team @OxfordIHTM/charmander, well done for all your work and contributions to the hackathon. You can now view the output report that contains your inputs from the script that you submitted a pull request on here - https://oxford-ihtm.io/ihtm-hackathon-2023/coverage_assessment_report.html

I have given direct feedback to the group and to @quyentrano for the pull request that you have made. In addition to that, I am giving specific team feedback based on your efforts as a team during the hackathon.

I am impressed with the use of the {gtsummary} to create the output tables for the results. That is a great find and indicates willingness to go out of the box and find answers outside of what has been taught. I hope the rest of the team will look into how that package was used so that you can learn how to use it as well.
Your team is probably the one team that was consistently mostly in attendance for the three in person hackathon sessions. Thank you for taking the effort to be in class during these sessions.
Your team was also the only team where each member went out of their way to make their own RScript rather than code within the section of the already existing RScript in the project. This shows initiative and willingness to try something on your own.
On the flipside, one of the things that we were trying to emulate in the hackathon was about learning how to code with other people. Whilst having a separate RScript is not wholly detrimental, completion of a full workflow became more difficult. You can see how the full workflow looks like with your contributions in it by pulling the latest commits and merges from your RStudio and then from the main branch, you will see the coverage_assessment.R script/workflow that contains all the contributions from the different groups. All three other groups were able to make pull requests with their code within the sections of the main R file and this made putting things together much easier. So, I think key reminder here is that your individual efforts are important and are appreciated (if you can see, your separate R Script is included in the final and official output repository/project. But it would be good to also take into account what the entire class is trying to achieve with a full report on all sections of the analysis.
From both the coding and statistics side of things, important for you to remember the notion/concept of NA in R. NA is a special value and it is neither a character value nor a is it equal to 0. When we represent NA in R, it should be stated as NA and not as "NA". Doing the latter has implications in how basic statistical measures are calculated (mean, max, min, sd) and the sooner you appreciate this, the better because NA is a constant feature of any dataset and there is no way getting around that. If you have the time, I would recommend you looking at your submitted code again and re-check the impact of using "NA" instead of NA.
Whilst the use of the {gtsummary} package is great, one limitation I found specific to your team is that the output tables are more presentational and if you need to access the actual values for other purposes, you didn't have the actual table of values available to you. In reports, this is useful because when writing the text in R Markdown, you can write code that will provide you the actual value within the text and if for some reason the values change, you don't have to edit/re-type the new results as the code will updated based on your new results. You will see how we do this in the outputs for the other groups.

Well done again to your team! Thank you for all your contributions! Look forward to the course dinner in late April as we will be handing out awards and tokens for your contributions.

OxfordIHTM / ihtm-hackathon-2023