I worked at Royal Free Hospital for one year. When the 1st National Friends and Family test was published at the end of July 2013, I was surprised that Royal Free Hospital had the lowest net promoter score. I have worked in many hospitals and I actually thought that patients were treated reasonably well. In fact in the 2012 National Staff Survey when asked: ‘If a friend or relative needed treatment I would be happy with the standard of care provided by this organisation’, 72.5% answered ‘Agree/ Strongly Agree’ which was in the 4th quartile of performance.
So I asked myself a question, is there are correlation between the ‘Friends and Family test’ and the ‘NHS Staff Survey’?
I took both datasets, combined them and ran a linear regression analysis.
As you can see, the R-squared value is 0.1513 which represents a small positive correlation. I’m afraid that my statistics is not up to scratch to ask if this is statistically significant.
Interestingly, my old hospital does look to be an outlier.
I don’t know what this means but a few things but there are a few things to keep in mind
- The response rate for the friends and family tests for inpatients was only 10% at Royal Free Hospital
- The net promoter score is calculated very differently from the NHS Staff Survey (please see links above for an explanation).
From the website:
Those that say they are ‘extremely likely’ are counted as promoters. ‘Likely’ is neutral, ‘neither unlikely nor likely’, ‘unlikely’ and ‘extremely unlikely’ are all counted as detractors.
I welcome others to analyse this data. The sources and methods on how the raw data was collected and calculated are available from the sources (links above).
Update: 0734 19/08/2013
This post has generated quite a bit of interest with a lot of talk about sample sizes and response rates.
I wonder if someone able to create dynamic graph where the user can filter by the sample size and when mouse-overed, the tooltip will show the hospital the data pointrefers too? Maybe the size of the data point can correspond to sample size too (like a bubble graph?). All help greatly appreciated.