This site is intended for health professionals only

Data – and how to get what you need

With GP commissioning now centre stage, good quality data is more important than ever. Emma Wilkinson asks three experts how GPs can improve their use of data.

What do you need data for in commissioning?

SF If we're going to change services we need to know what's happening at the moment, the numbers going through the system and how much they cost. We also need to know about referrals by practices and elective admissions to assess where the opportunities are and where something can be done about it.

AH GPs need to start with the data and use that as a base for building up a full picture of what's really going on. The data on its own is an indicator but to get the whole story you need to add other things otherwise there is a danger you are trying to tackle problems that don't exist.

One of the things I try to bear in mind when I'm finding information is that numbers are just one part of the picture. Where we have gone wrong in the past is jumping to a conclusion about what the data is telling us about admissions, for example. But on its own that is not enough. You need additional information to put it in context.

TD Data raises awareness of where your money is going and where your referrals are going. It gives you an insight into what the contracts mean for foundation trusts as well as for the practice. It allows you to look at what you're doing and what you could do better.

With your own practice data you can look at referrals, whether there are any discrepancies between the partners and whether everyone is making full use of the facilities within the practice. For example, if you have a doctor who does joint injections, are patients being referred there or to the hospital? With PCT data you can see how many patients are being seen and the cost of their appointments.

Why do we keep hearing people moan about the quality of data in commissioning?

SF For many areas, people have simply not scrutinised the data coming out of secondary care in as much detail as they should have.

In the days of fundholding GPs looked at every single invoice and tallied that with their own data. Now coding is much more complex and although many foundation trusts are already very good at coding up procedures, we also know that a certain proportion of the stuff that comes out is just wrong.

There is huge scope for practices to look in detail at that data to make sure they're paying only for services that patients have received but also to drive up the quality of data. We're going to have to do a lot more work on data from our mental health trusts and community providers.

AH Quality has become far more important suddenly, and GPs have started to become concerned about whether the quality is suitable for the job.

Small errors that might not have been important in huge datasets suddenly make a big difference when you're looking at small practice-level data.

There are lots of quality issues and different areas will have different problems. A&E data can be poor because it is from the front line where people are in a bit of a state and might not remember who their GP is, for example.

Then we have people coding data that doesn't always fit neatly into a category so it can be ambiguous. A lot of work is needed to convince GPs about the quality of the data. To some extent they have a right to complain but sometimes they misunderstand what the data is.

TD Quality can depend on how your PCT and trust works but I suspect that most trusts are getting much better. I hope that at a practice level the data is being recorded accurately.

Data from mental health and community teams still has a long way to go but in the future it will be in their interests to associate costs with patients because they will need to show who provides most value for money. One issue for trusts is that it depends who codes the data. If it's a ward clerk, they might not understand the significance of their coding in terms of finances. It's important to remember that it can be contractual now for trusts to provide timely data – so this can be addressed if it is an issue.

Why is data so important for GP commissioners?

SF Where GPs have started to look at data they have also started to make significant savings. But there is an issue about the sort of data we should be collecting at practice level and how that could change in the future.

I would like to see a system where, if a GP decides to refer, the system will tell them the implications of that referral in real time like it does with prescribing. This should be relatively easy because in fundholding we had an accounting system that accrued costs of referrals and so you knew where you were from month to month.

AH Although we have been working on this for ages, one of the issues that comes up is that the information often isn't what GPs want or hasn't been that useful. So this is a good point to start again. GPs are often very interested in the data if it has been presented the right way. Although data is vital, just handing GPs an Excel spreadsheet will not work. If you can present data well and make the message clear, whatever that message is, it will be something they can use to improve services.

TD If the white paper is taken to its logical conclusion, GP commissioning is going to have a huge impact on the future income of practices and by looking at the data we can see what contract to work in. For example, if a PCT has negotiated a contract for 500 orthopaedic referrals but only 300 are seen, that points you in the right direction. The more information you have the more efficiently you can commission services and anticipate changes and future needs.

What are the sources of data?

SF There is obviously hospital and practice data and as I've mentioned we're going to need more information from community nurses and health visitors, because we can do better there. There is also a lot of public health data, which is not too bad but does need to be far more localised. Data on morbidity and mortality needs to be available by cluster rather than by PCT or even county if it is going to be useful.

AH One thing to point out is that hospital datasets aren't going to be particularly useful for GPs in the future. We need to be able to bring together clinical outcomes with patient outcomes. For example, for the patient, the outcome of a hip replacement isn't clear until they come home. We need to know if the operation was successful from the patient's point of view so the data we have now is not adequate for that.

TD As well as practice-level data, PCT data, and medicines management data from the PPA, there are national statistics and health needs assessments. Maybe now we should even be looking at local authority data and their requirements to combat things like fuel poverty.

Does data have to be 100% perfect to commission a service?

SF It doesn't have to be perfect for GPs to make broader changes. But if GPs are going to start to move services from secondary to primary care they need to know, in detail, what's happening. To do this properly we will need very good patient-level data.

AH It will never be perfect. It's a reflection of what's happening. What's important is that people agree that the dataset is good enough for their purposes. It's about identifying basic datasets that are accurate and working from that point.

Some practices are ahead of the game, so everyone is starting from a different place and that is what might be difficult. It's about trying to make GPs feel well enough informed to make decisions and remembering that's what the data is there for rather than to bully people into certain actions. Really we are all trying to solve the same problem. The best way of convincing GPs that the data is sound is to try to validate it with their own practice data.

TD I don't think the data has to be perfect. It has to be reasonably accurate, but you're always going to get one or two patients who have moved, for example.

And you want to be thinking in the long term. As long as you have a picture in your mind, it may be that in the future the way contracts are commissioned wouldn't be down to each patient. Maybe we'll be looking at groups of 100 patients. Also practice managers have an awful lot on their plates to be chasing up patients at the micro level. We are going to have to wait and see how important data at the individual patient level is going to be.

How do you set a suitable benchmark of quality given the limitations?

SF If you're going to be paying only for data that's accurate you shouldn't have to worry about benchmarking. If providers have poor data they won't get paid so the onus will be on them to be accurate.

AH If the data is wrong you should go back to the acute trust for discussion – it's about correcting errors. People rarely deliberately manipulate the figures.

TD I don't know how you would benchmark quality other than to have quality in terms of substance.

For instance, if a diabetes patient has had a knee replacement, is it more important to make sure the knee replacement is the primary diagnosis or is it more important to get the discharge letter the day after discharge and check that all the medication is correct?

We must establish what we want from quality and I suspect the discharge letter is a lot more important. Timeliness and quality of information are the sort of thing general practice can commission for. You need a wish list of quality issues you can put in the contract.

Emma Wilkinson is a freelance journalist

Dr Stewart Findlay Dr Stewart Findlay data experts Potential sources of data for commissioning

For more detail see

• Area profiles give an overview of your area and how it compares with others.
• Child and Mental Health Observatory (ChiMat) includes online tools to support commissioning.
• Clinical and Health Outcomes Knowledge Base provides comparative data for 700 health and local government organisations in England.
• Connecting for Health Secondary Uses Service (SUS) for commissioning datasets on NHS-commissioned admitted patient care, outpatient and A&E activity.
• Department of Health website – statistics page for reports and commissioning tools.
• Dr Foster Intelligence online tools for tracking performance, formulating strategies, monitoring implementation and tracking improvements.
• Electronic Prescribing & Financial Information for Practices (ePFIP) for comparing prescribing performance against national and PCT comparators.
• Health profiles from Association of Public Health Observatories report the health of people in each local authority and highlight inequalities.
• Health and Social Care Information Centre for statistics.
• Hospital episode statistics data on NHS patients.
• The Information Centre for national comparative data for secondary uses.
• NHS Comparators compares local commissioning activity, costs and outcomes.
• Office for National Statistics provides health statistics on a national basis including census data.
• Primis + a free service to aid effective use of clinical computer systems in primary care.
• Practice data for checking referrals and verifying secondary care information.
• Quality and outcomes framework database for GP data going back to 2004/05.

Source: NHS Evidence – commissioning