We need to know more about what happens in general practice

With widespread hope that the new general practice dataset will develop into a comprehensive and robust resource, Camille Oung assesses the dataset as it stands and suggests ways progress might be made to bring the information up to the levels seen elsewhere in the health care system.

Blog post

Published: 26/06/2019

General practice is held up as the ‘bedrock of the NHS’, and leaders tell us that if ‘general practice fails, the whole NHS fails’.

Supporting and improving these services is therefore central to the 10-year vision for the NHS set out by this year’s Long Term Plan. It was accompanied by a new GP contract, packed with commitments to boost funding, reorganise the workforce, and provide innovative care through an increased use of online or technological platforms.

Yet compared to most other parts of the NHS – particularly secondary care – we know remarkably little about what actually goes on in GP surgeries. The sparsity of routine statistics about activity (let alone outcomes) makes it very difficult to identify the right way forward. With signs of strain evident as recent patient surveys report satisfaction at an all-time low, it’s become more important than ever to get an idea of the effectiveness of activity in general practice.

In hospitals, routine datasets such as Hospital Episode Statistics (HES) give us insight into who gets treated, for what, by whom, and where. This basic knowledge of the activities and performance of hospitals is now essential for making evidence-based policy changes, monitoring existing policies and providing fundamental administrative data to run the system and allocate funds. There is a clear need for something similar in general practice – but since the demise of care.data, a HES-style data system for general practice – progress has been slow.

The GP appointments dataset: a watershed for primary care?

In December 2018, NHS Digital released a new open-access dataset containing information fed in from practice IT systems for over 300 million appointments in general practice from November 2017. This is a step in the right direction. Its monthly updates constantly improve the data, bringing in ever more practices: by March 2019, 91.5% of practices across England had been included.

Some interesting insights have come from the dataset already. NHS Digital’s Dashboard provides an overview of indicators at a national and CCG level, showing an appointment’s status, mode of delivery, and the type of health care professional that delivered it. Others have used the data to ask some questions about appointment delivery: which days of the week are busiest? Is there a relationship between the length of delay from booking to appointment and non-attendance?

In time, the dataset could help us answer important policy questions: for example, the effect of the GP contract on continuity of care as patients gain more freedom to choose the type of health care professional they see at the clinic.

But we’re not there yet

NHS Digital makes it clear that what has been released so far still needs improvement. As it stands, there are a number of problems with the data and these limit the extent to which we can use it as a resource for evidence-based policy.

Having different practice IT systems gives GPs flexibility over the way they organise their work. But the absence of reporting standards in general practice means information is recorded differently in each system, making it difficult to combine the data, and information gets lost. This is most obvious with the reporting of health care professional type, where the use of different categories means that the dataset only distinguishes between ‘GP’ and ‘Other practice staff’.

There is currently a strong drive for team-based approaches in general practice, for example by making greater use of pharmacists and physiotherapists. How can we tell whether these work when we cannot tell a dispenser from a district nurse, from an interpreter, or from an osteopath? In addition, including new practices tends to result in jumps in the proportion of different professions, making it difficult to tell what is really going on – for now at least. The figure below shows how these broad categories have varied in proportion as new practices with different reporting patterns are brought in.

Proportion of appointments by health care professional type from January 2018 to April 2019 21/06/2019



November and December 2017 were excluded from this analysis as the all health care professional types were reported as ‘HCP Type not provided’.

Read more

The lack of consistent reporting standards across multiple providers also means blips in one system can hinder the accuracy of the entire dataset – which perhaps explains why it’s taken so long to see something like this. For example, large chunks of the information on appointment attendance are unusable from November 2017 to October 2018 owing to an error in capturing ‘Did not attend’ statuses in practices using the TPP system. For these months, as the figure below shows, the data implies that not one single patient attended an appointment.

We frequently investigate the association between waiting times, attendance rates, and emergency admissions. It’s a shame that almost a year of reporting can’t be drawn upon in this context.

Proportion of reported appointment rates from March 2018 to March 2019 21/06/2019


Read more

Like for like: bringing GP data up to par with hospitals

The appointments dataset gives us the starting point and the opportunity to discuss the evidence we really need to see in general practice. So what needs to change to make the data on activity in general practice as useful as possible?

Our experience with HES offers some valuable lessons in dataset development. Although HES is not without flaws, it has gained in strength and value over the years through constant collaborative improvement efforts. Consistent guidelines for reporting and cleaning the data have made it a reliable resource used for countless purposes to understand, monitor and improve the NHS. It’s good to see the recent Future IT GP framework make a promise that this will happen for general practice too.

Furthermore, the types of analysis that HES makes possible show us how GP data can be more comprehensive. For example, with concerns about inequality in health care access, it would be useful to be able to know things such as the area in which a patient lives, as we can in HES, so that we could match it to data on demographics and deprivation. Including clinical information can also help us discriminate between good management of conditions and inefficient practice activities.

As a research and policy community it is up to us to ensure the GP appointments dataset receives the attention it deserves. Otherwise, the data we have makes it all too easy to focus on patients only once they get admitted. We’ve spent long enough talking about the importance of what happens in general practice: it’s time we were able to see it.

Suggested citation

Oung C (2019) "We need to know more about what happens in general practice”, Nuffield Trust comment.