Whether you are a Chief Executive, Operational Manager or part of the Clinical and Nursing team, analyzing and interpreting data will already be a vital part of your job. Data can be represented in various forms from the latest board pack to dashboards displaying patient demographics. Regardless of the style and format, the ability to understand what the data is telling you and identify trends will be essential for decision making.
Where did that chart come from?
Before you begin analyzing a report, consider the source of the data, how the data was measured and whether the author of the chart has placed particular emphasis on a point or fact to highlight a specific issue. For example, trusts reporting the number of patients waiting more than 12-hours in A&E use the time of arrival, whereas NHS England are currently reporting the “trolley wait” figures [1]. Both measures are valid, but the “trolley wait” figures are significantly lower as the measurement is different. Being aware of how data is captured and measured will strengthen your understanding of what is being reported and enable you to challenge the data displayed in reports.
Dashboard Overload
Dashboards offer a great visual representation of an organization's performance, but the number of charts displayed can be overwhelming at first glance. Start by focusing on one chart in the dashboard and maximizing it. If you are using Microsoft Power BI click on the chart and select Focus Mode. Given the limited space available in dashboards, even the titles for the vertical and horizontal axis might be omitted so check the chart title again. For example, where the chart has the following title: ‘A&E Attendances Seen for Treatment/Left Before Treatment (Rolling 6 months)’, we can assume that the x-axis is displaying data by month and the y-axis the number of A&E attendances.
Scales start at 0 (most of the time)
When designing basic charts the normal convention is to start the vertical axis at 0 [2], using a linear scale as shown in Chart 2 below.
Using the same data as in Chart 2, the scaling in Chart 3 has been modified so that the y-axis starts at 100k rather than 0. Although the change in the number of attendances each month is much more evident due to the height of each bar, the difference between month 3 with 105,697 attendances (highest) and month 7 with 101,195 attendances (lowest) is only 4502 attendances. Adding data labels has helped but note that your chart may not include them so always check the scale.
Averages
Chart 3 includes an average line displaying the mean number of typical attendances calculated using the sum of all attendances divided by the total number of months. Using the mean value makes sense as the number of attendances does not vary greatly each month.
Where a data point differs significantly from the other data points in a chart (outlier), the median provides another way of calculating the average. Take for example pay scales in the NHS, where the entry salary for Band 1 starts at £18,870 with Band 9 £93,735 (ignoring mid and top step points) [3]. The mean salary is £43,177 but the median (middle value of the pay scale) is £28,981. The median represents a more realistic figure for the majority of staff working in the NHS.
Mode is another type of average representing the most commonly occurring value in a set of numbers. In Table 1 below, the length of stay (LOS) with the highest number of patients is 4 therefore the modal value is 4.
Note that averages are not mutually exclusive, and charts may include one, two or all three averages depending on the data. Just remember to check the type of average displayed.
Useful Tip: If you are still struggling to interpret the data in a dashboard using Microsoft Power BI, try copying the chart and changing the visualisation to ‘Smart Narrative’. Power BI will summarise the chart and present the highest and lowest values, percentage, average values and range in a text box.
Value of Data
So finally, why the focus on data? NHS data is reportedly worth £9.6bn [4] in improved patient outcomes, operational savings and benefits to the economy as a whole. However, the volume and complexity of data makes analysis even more difficult despite the advent of multiple data visualisation tools. By reading this blog and analyzing data with the above points in mind, I hope that those of you who are not Mathematicians or Statisticians have a better chance of understanding what the data is telling you and will improve your ability to make timely and informed decisions.
[1] HSJ (2022) Available at: Trusts reveal thousands of new 12-hour waits | News | Health Service Journal (hsj.co.uk) (Accessed 2 August 2022)
[2] Open University Available at: Working with charts, graphs and tables: 4.4 Bar charts - OpenLearn - Open University (Accessed 27th July 2022)
[3] NHS Employers (2022) Available at: Pay scale material for 2022/23 | NHS Employers (Accessed 27th July 2022)
[4] digitalhealth (2019) Available at: NHS data worth £9.6bn per year, says Ernst & Young (digitalhealth.net) (Accessed 27 July 2022)
