A few months ago, I was speaking to one of the Health Foundation’s GenerationQ fellows, a GP working to improve the care of her patients. She mentioned that she and her colleagues had been reviewing the reasons for their patients’ emergency admissions and found a surprising number were admitted because of pneumonia.
However, when they explored further, they found this was often not the case. The actual reason was frequently something else – a fall, for example – but the patient either contracted or was diagnosed with pneumonia while in hospital, and this was recorded in the discharge summary as the reason for admission.
We recently published a study of Hospital Episode Statistics data that found another, related, issue: that patients’ long-term conditions are often not recorded in hospital data. Congestive heart failure was only recorded in 53% of admissions where it was not the main reason for the patient’s admission, while hemiplegia/paraplegia (paralysis) was only recorded in 43% of cases. Of the six long-term conditions we looked at, diabetes was the most consistently recorded, but still recorded in only 89% of cases. This was despite NHS coding guidelines, which state that these long-term conditions are clinically relevant and should always be recorded. These two examples show that there is room for improvement in the quality of hospital data.
The uses of hospital data
As part of the day-to-day operation of NHS hospitals, massive datasets are routinely generated. These contain information on every A&E attendance, outpatient appointment, emergency and elective admission in every NHS hospital in England, including information on treatments and patients’ long-term conditions. There are not many other databases in the world with such breadth and wealth of health care data.
These data can be and are used for a wide variety of purposes. Of course, they are used to support the care of the patient while in hospital, as the basis for payments to hospitals, and by CCGs and NHS England to track and plan services.
But its uses go beyond this. It allows GPs to tailor care to their patients. Analysts can also use it to examine the success of an intervention in improving patient outcomes, or investigate and quantify variations in the quality of care or need. I remember a few years ago a QualityWatch report that revealed that people with mental ill health were admitted to hospital in an emergency for physical needs nearly four times more often than other people. Research such as this can help inform and shape both local and national policy decisions.
Why does the quality of hospital data matter?
Regardless of how the data is used, the better the data, the more valuable it is. Good-quality data not only helps patients receive better care, it makes for better research and analysis too. Although we, as analysts, can be thoughtful on how we use the data – by taking its limitations into account, or working around them where possible – the quality of the insights, and the accuracy and reliability of the conclusions we can draw, can only be as good as the quality of the data.
High-quality data are also important for innovation. The Government has recognised in its industrial strategy that data are an important asset for the UK that could be used to develop new tools, such as algorithms, to diagnose and treat illness earlier. Again, better-quality data means the data are more valuable.
How can we improve the quality of the data?
There are several steps between a patient being diagnosed and the corresponding data being recorded in the hospital records. Each step is important in ensuring the quality of the data, from the health care professional retrieving information from previous admissions, to the writing of medical notes and discharge letters, and translating these into standardised diagnosis and procedure codes by clinical coders.
These steps ideally require good IT systems, allowing easy access to reliable and complete information, which is not always the case. Often, information collected in A&E is stored on a different computer system from that used on wards, and only information on previous admissions to the same hospital is available within that hospital. Some steps to improve IT are outlined in the NHS Long Term Plan, for example the expansion of the Global Digital Exemplar programme.
However, as the GP example shows, we also need a common understanding of what information it is important to capture and why, a consistent approach to recording this information, dedicated time for doing so, and – maybe most important of all – a general understanding and acknowledgement at all levels within the NHS of the value and importance of its data. Without this, both the quality of the data and the insights that it can provide will remain limited.
Thank you to Susannah Pye, Paediatric Registrar, for her insight into hospital processes.
Therese Lloyd is a Senior Statistician within the Improvement Analytics Unit at the Health Foundation