Date: 21 March 2017
Category: GP patient survey
As the statistician on the IMPROVE study team, it was my job to approach data with scepticism. I had to think constantly about how and why the data we had generated, and more importantly the national GP Patient Survey data, might mislead our readers and how we could best examine the data to shed light on it. I had to be sure we were conveying the right ideas about ways to improve patient experience.
Whilst it has its problems, overall I’m reassured about the data at the heart of the IMPROVE study, the GPPS data generated by respondents to the annual survey. I really want people to realise just how robust the data is.
Ipsos MORI, who run the survey on behalf of NHS England, put a lot of effort into coming up with a good sample and making the data fair and representative. We know that many people who read the GPPS results for their practice are sceptical about them. They say “the survey isn’t accurate, it’s biased because it only gets answered by people who have got an axe to grind.”
In fact, the data suggest that as far as doctor and nurse consultations are concerned, people are more positive in the GPPS questionnaire that they should be because, as we established, they are reluctant to criticise their GPs. I’d like practices to study the data and to think: “well, OK, our consultation ratings are 85% and that looks great but maybe that is overstated and so there is room for improvement.” And I’d really like practices that serve predominantly South Asian populations to consider what we found out about the experience of South Asian patients in comparison with their White British counterparts and to think about what could be done to address that in their own practice.
So, if you work in a practice, don’t expend a lot of effort doing your own local surveys because you doubt the value of the GPPs data. The GPPS data gives you good insight into how you are performing compared to your peers and it is rich – dig deeper into it!
Statistician Gary Abel discusses the credibility and robustness of data produced by the GP patient survey.
IMPROVE Research Project Statistician