Our first task in Dashboard Week was one that also involved the use of APIs, specifically, an API from the Australian Institute of Health and Welfare with a whole variety of information from hospitals all across Australia.

The one piece of advice that we were all given was to learn how to fail fast and timebox efficiently, which meant coming up with something to focus on and then trying to find any type of possible insights afterwards. A quick browse of the available datasets in Postman showed a few categorised under “Hand Hygiene” which intrigued me just from the name.

I built out this workflow in Alteryx  to download the 78 total datasets, 19 on hand hygiene ratings and 19 on counts of hand hygiene moments in every hospital in Australia. I had to create a simple batch macro to achieve this, and fed it the dataset IDs of everything I wanted before continuing on in Tableau. The one important thing to note here is that the date of when the data was created is only in the title of the dataset, so I had to add an extra column for the year before exporting.

Once in Tableau, I identified that I wanted to only focus on New South Wales as I lived here and also that many of the more remote hospitals in other states had caveats for their rows, either due to having less than 100 moments of hand hygiene being recorded (most likely due to the size of these facilities), or have a lack of acute hospital beds, resulting in nulls in the value column. After that was done, I created BANs for the last year of data – 2022’s average rate of hand hygiene in NSW hospitals and the number of hand hygiene moments, as well as the percentage difference from the previous year. This was actually harder than I anticipated since a member of my cohort, Prerana, usually handled these types of visualisations in the client project weeks so it was a good opportunity to learn.

I threw up some line charts on the change of average rate and average count of moments over time, as well as a bottom 10 bar chart of hospitals by the rate of hand hygiene. The government strives for a rate of 80% hand hygiene in Australian hospitals as the benchmark to meet, and I was pleasantly surprised that only the two worst performing hospitals in this area failed to meet that requirement and only by a few percent.

I also brought over spatial data from the API to plot all the hospitals on a map as a way to show the highest concentration of moments, which was unsurprisingly almost all around the Sydney area. Had I more time I would’ve liked to make the BANs more dynamic to the date filter, and find a way to union the data sets better.

Daniel Yam
Author: Daniel Yam